Shreshtha Priya
Student at Step By Step School, Noida, applying to study civil and structural engineering.
- Structural resilience
- Sustainable materials
- Disaster-resilient design
About me
I'm Shreshtha, an IB Diploma student from Noida, and I want to study civil and structural engineering. Away from engineering, I love theatre, basketball and music.
I have always loved building things. It started with Lego and Hot Wheels, and over the years it grew from toys into real materials and real structures. Today that is Eco Brick, and most of my work is about building materials and the people who depend on them when a flood comes.
My grandfather's village in Puri, Odisha floods almost every monsoon. That is where Eco Brick began. Since then I have tested bricks made from plastic waste and scrap metal, built granaries with village masons in three states, designed a space settlement structure with my school team, designed a boat that cleans drains, and built an AI research system with a postdoctoral researcher at MIT.
Each project below opens to its full story.
Projects
What I've worked on
Hover over a petal to read about a project. Click it to open the project.
Open projectSolo research
Eco Brick
A hybrid interlocking brick made from recycled plastic, cement and scrap metal, built for villages that flood every monsoon.
- Role
- Designer and researcher
- Material
- Plastic, cement and metal composite
- Geometry
- Tri-axial, one piece
- In use
- 5 granaries, 3 states
How it started
As a child I spent my holidays in my Naanu's village, Beleshwar Patna, in Puri district, Odisha. The walls of the houses there had long black stains running across them. I used to think they were sunburns.
They were flood marks. Every monsoon the water rose into the village and stayed for days. When it went down, it left a line on every wall it had touched. Most houses had mud walls. Heavy rain weakened them each year, and many collapsed. Families rebuilt them after every monsoon, knowing the water would come back.
Years later I started working on a brick that could stand in that water. I wanted it to cost little, last longer than the fired clay bricks used in the area, and be simple enough for local masons to build with. That brick became Eco Brick.
The problem
Around 40 million hectares of India, more than a tenth of the country, is prone to floods. The worst affected areas lie along the Ganga and Brahmaputra basins and in coastal states such as Odisha, where floods come almost every year. Families there mostly live and store food in mud and fired clay structures, which weaken in heavy rain and often fail when the water rises.
About 10% of India's food grain is lost after harvest, and around 6% of those losses come from poor storage. Small farmers, who make up over 80% of the country's farmers, lose the most, because they keep grain in traditional stores that give little protection from moisture. For a family living on one harvest a year, a single flood can ruin months of food.
Sources. Flood-prone area from the Rashtriya Barh Ayog and the National Disaster Management Authority. Post-harvest losses from the Journal of Postharvest Technology and Food Control (2025). Share of small farmers from the National Economic Forum (2025).
The brick
The brick at a glance
Strength
Thin metal inserts carry load inside the brick. The final brick showed 68% higher compressive strength than conventional concrete bricks.
Ductility and crack control
Plastic strips let the brick deform before it fails and slow the spread of cracks.
Low water absorption
A waterproofing hardener and the tri-axial shape cut water absorption by a third against an unreinforced brick.
Tri-axial Eco Brick
Cross-section, illustrative
Minimal mortar
Each brick locks into its neighbours along three axes, so a wall needs only minimal mortar and little skilled labour.
Impact absorption
The brick absorbed over 90% of impact energy, about 42% more than conventional concrete.
Fire stability
Every specimen stayed intact through more than 30 minutes of direct flame.
Why a hybrid
Most building materials trade strength for ductility. Cement is stiff and strong but cracks suddenly. Plastic bends but is weak. Eco Brick puts metal inserts in for load transfer and plastic strips in for crack control, so that one material does not have to do both jobs. Recycled plastic and scrap metal are cheap and easy to find, which keeps the cost of each brick low.
Making it
How Eco Brick is made
Collection and cleaning
Plastic waste is collected and cleaned.
Shredding
The plastic is shredded into pieces of even size. Early batches with uneven pieces mixed badly, so I added size control here.
Mixing
Sand and Ordinary Portland Cement are mixed 4 to 1 with the shredded plastic, water and a waterproofing hardener (Dr. Fixit Pidiproof LW+).
Reinforcement
Plastic strips and thin metal inserts are set into the mix.
Moulding
The mix is cast into a rectangular, bow-tie or tri-axial mould.
Curing
Specimens cure in ambient conditions for [number] days.
Testing
Each specimen is tested for compression, impact, water absorption and fire.
I made four mixes for every geometry so that each type of reinforcement could be judged on its own. The quantities are for one specimen.
| Brick type | Sand | Cement | Water | Plastic strips | Metal inserts |
|---|---|---|---|---|---|
| Unreinforced | 400 g | 100 g | 40 to 60 ml | none | none |
| Plastic | 400 g | 100 g | 40 to 50 ml | 25 to 30 g | none |
| Metal | 400 g | 100 g | 40 to 50 ml | none | 40 to 50 g |
| Plastic and metal | 400 g | 100 g | 40 to 50 ml | 25 to 30 g | 40 to 50 g |
Geometry
Three shapes
I tested three shapes with every mix. The rectangular brick behaved like a normal brick, and cracks ran straight along the line of the load. The bow-tie spread some of the stress along its diagonal edges. The tri-axial brick is one piece with three arms set at 120 degrees. It performed best under both compression and impact, so I chose it as the final design.
How the interlock works
The notches between the arms of one brick take in the arms of the bricks around it. Every brick is held on all sides and locked along three axes, which raises the contact area between bricks. When the wall takes a hit, the load passes into the neighbouring bricks and spreads over a larger area, so the stress at any one point drops.
Testing
How it was tested
Every shape and mix went through the same four tests under the same conditions, so the results could be compared directly. I built the compression and impact rig myself around a linear actuator.
Compression
The actuator pressed down on each specimen until it failed.
P is the failure load and A the loaded area.
Impact
A sudden load struck each brick to measure how much energy it absorbed.
m is the striker's mass and v its speed.
Water absorption
Specimens were weighed dry, soaked and weighed again.
Fire
Specimens sat in direct flame for more than 30 minutes and were checked for cracking, breaking and interlock failure.
Verification, 20 January 2026
Universal Testing Machine
I then had 13 specimens tested for compressive strength on a Universal Testing Machine at ARC Infinity Lab Private Limited in Mumbai, under IS 516 (2021). Among the tri-axial bricks, the hybrid with both plastic and metal was the strongest of the four mixes, at 6.8 N/mm². The test confirmed most of my earlier results.
Results
How the final brick compared
No specimen of any shape broke apart in the fire test. They changed colour and showed light charring, and every interlock held.
Sustainable Development Goals
Young SDG Leaders Program
In December 2025 I took part in the Young SDG Leaders Program run by InnovaPolis. There I learned about the 17 Sustainable Development Goals and the specific targets that sit under each one.
I also learned how a project can be adapted to address more of the goals. I used that to look again at Eco Brick and the granaries, and to map the project against the five goals below.
Zero Hunger
The granaries keep stored grain dry through floods. Target 2.4 asks for farming systems that can adapt to extreme weather and flooding.
Industry, Innovation and Infrastructure
Eco Brick is a new material for rural building. Target 9.1 asks for reliable, resilient infrastructure that people can afford.
Sustainable Cities and Communities
Buildings that stand through floods reduce what families lose. Target 11.5 aims to cut losses from water-related disasters.
Responsible Consumption and Production
The brick reuses plastic and scrap metal (Target 12.5), and the granaries reduce post-harvest food loss (Target 12.3).
Climate Action
Floods in India are getting more frequent. Target 13.1 aims to help communities withstand and recover from climate disasters.
Recognition
British Science Association
CREST Gold Award
The project received a CREST Gold Award in April 2026. CREST Gold is given for long, independent research projects.
January 2026
IRIS National Fair
I presented the project in the Materials Science category at the IRIS National Fair, held from 29 to 31 January 2026. The project placed in the top 75 of more than 2,100 entries.
Physics, team Phy-153
S.T. Yau High School Science Award (Asia) 2026
My research outline on the brick qualified through the first round, and I was selected for Round 2.
September 2026, Bangkok
ISBM 2026 conference paper
A paper I co-authored on the brick was presented at the 5th World Conference on Information Systems for Business Management and selected for publication in its proceedings.
Impact of Eco Brick
Granaries and Florit
Flood-resistant grain storage built with Eco Brick in three villages, paid for through a pilot project with Florit Construction.
- Granaries built
- 5
- States
- Bihar, Jharkhand, Odisha
- People reached
- 3,000+
- Approved next
- 10 granaries
In flood-prone villages, grain is usually kept in mud structures that let water in during the monsoon. A family that depends on one harvest can lose months of food in a single flood. The granaries are built with Eco Brick to keep that grain dry.
Each granary is built only after the village panchayat approves it, and the masons who build it come from the village. They know how high the water rises and where it settles, and that shaped where and how each granary was built.
In October 2026, Delhi's Minister of Public Works, Irrigation and Flood Control, Parvesh Sahib Singh, wrote a letter of appreciation for the project.
How it was funded
Pilot sanitation blocks, Varanasi
Florit Construction offered me a pilot project at its residential site near the Ganga in Varanasi, in a low-lying area that floods. I built six sanitation blocks with Eco Brick for the site's labour colony, which houses more than 100 construction workers.
I was honoured to be guided by Florit's chief engineer, who taught me about every part of the project as it was built.
Partner
Florit Construction
Design
Based on the FCI godown
The granaries are modelled on the conventional godowns, or grain warehouses, built by the Food Corporation of India (FCI), the government body responsible for storing India's public stock of food grain. These godowns are designed to keep large quantities of grain dry and free of pests over long periods. I retained the features that serve this purpose and scaled the building down to a size that village masons can build and a panchayat can manage. The walls use a mix of my Eco Bricks and conventional bricks, which balances the strength and water resistance of Eco Brick against the cost and local availability of standard bricks.
- Raised plinthThe storage floor sits 600 mm above ground level, which keeps the grain clear of shallow floodwater and ground moisture. FCI godowns follow the same principle, with a plinth of about 0.8 m.
- VentilationA continuous ventilator fitted with mild steel mesh runs high along the walls. Warm, moist air can escape, while the mesh keeps birds and pests out.
- Clear spanSteel roof trusses span the full 10 m width, so the 10 m by 16.8 m storage hall has no internal columns and grain bags can be stacked without obstruction.
- LoadingA ramp at a slope of 1 in 10 and a flight of steps rise to plinth level for loading and unloading.
- StructureThe walls combine Eco Brick and conventional brick, set within a reinforced concrete frame, with M30 concrete specified for the beams and slabs.
The granaries
I was also honoured with letters of appreciation from the mukhiyas and sarpanch of the three village panchayats.
Beleshwar Patna
Beldal Gram Panchayat. This is my grandfather's village, where the project began.
Pindra
Pindra Gram Panchayat, Mandu block, serving Pindra, Rauta and Tapin.
Sakir Bigha
Sakir Bigha Gram Panchayat, Belaganj block.
This cost-effective and sustainable technology completely protected the crops of our entire village during the floods.Sarpanch, Beldal Gram Panchayat, Puri
What's next
Ten more granaries have been approved in other villages. The project is also due to start in Delhi, in areas affected by Yamuna floods.
CCIR 1-on-1 Research Mentorship
Research with an MIT postdoc
A multi-agent AI system that makes its own atomistic data and trains machine learning models for materials science from one written instruction.
- Mentor
- Dr. Alireza Ghafarollahi, MIT
- Sessions
- Weekly, since June 2026
- Stack
- Python, AG2, ASE, MACE-MP-0
- Status
- Writing the paper
A machine learning model can predict how stiff an alloy will be, but only after someone gives it data. In materials science that data comes from slow experiments or costly simulations, and an expert has to guide every step. My project asks whether a team of AI agents can do the whole process, from making the data to testing the model, starting from one written instruction.
How the system works
A planner agent breaks the task into steps. A coder agent writes Python for each step. An executor runs the code and reports back. If the code fails, the coder reads the error and fixes it. Earlier systems of this kind could only call a fixed set of functions written in advance. In mine the coder writes new code each time, so the system can handle tasks nobody planned for.
The agents build alloy structures and run atomic simulations on them with MACE-MP-0, a machine learning interatomic potential. They then train and test a model on that data. I tested the system on copper-aluminium alloys, where it had to predict bulk modulus from only 50 samples in total. The final model reached R² = 0.986.
How the research progressed
Timeline
- June 2026
Learning to build agents
I set up a Python environment with Miniconda and Jupyter and called language models through the DeepSeek API. I learned AG2, Microsoft's framework for multi-agent systems, and built my first group of agents. I wrote my first tool, a simple calculator, and learned how type annotations and docstrings tell the model what a tool does. I then learned how agents pass control with handoffs, share state through context variables, and return results with ReplyResult. Between sessions I read ProtAgents, a paper from my mentor's group.
- July 2026
Atomistic simulation
I moved to the physics. I learned how to set up a simulation cell in ASE with periodic boundary conditions, and why the cell must be large enough that atoms do not interact with their own copies. I found an element's lattice constant by building structures at different spacings and taking the one with the lowest energy on the energy against distance curve. I studied FCC and BCC crystals, where FCC metals such as copper and aluminium are ductile and BCC metals such as tungsten are strong but brittle. I also learned how interatomic potentials work, from Lennard-Jones to machine learning potentials trained on DFT data from the Materials Project, and the difference between molecular statics and molecular dynamics.
My task was an agent that takes an element and computes its lattice constant by calling an ASE tool. I built it and then extended it to bulk modulus.
- August 2026
Elastic properties and alloys
I studied Bayesian optimisation and active learning, where a surrogate model maps composition to property and the search balances exploration against exploitation. I learned how a material's stiffness is described by the elastic tensor. Its 36 entries reduce to 21 independent constants, and for a cubic crystal like copper only three remain, C11, C12 and C44. My agents could now compute the elastic constants of copper when asked. I then looked at alloys, including phase stability between FCC and BCC, phase diagrams, building an alloy by swapping atoms at random positions, and the Embedded Atom Model.
B = V ∂2E∂V2B = C11 + 2C123A = 2C44C11 − C12Bulk modulus from how energy changes with volume, bulk modulus of a cubic crystal from its elastic constants, and the anisotropy factor, which equals 1 for an isotropic material.
- September 2026
Rebuilding the system
I restructured the system around AG2 patterns in place of direct chats, with context variables carrying state between agents. I rewrote each agent's prompt into clear sections with a fixed output format, gave the coder a shared library of alloy tools, and made errors explicit. The executor now ends the run when the code works and sends it back to the coder when it fails. All data is saved as CSV, and a summary agent reports what happened in each run.
- October 2026
Testing and writing
I ran the full system on copper-aluminium alloys and started writing the paper in Overleaf. Its subject is how the reasoning of language models can be combined with agent systems for faster materials discovery.
What I learned
AI agents need firm structure. I learned to keep routing, checking and error handling in ordinary code, and to let the model write only the parts that need to be flexible.
I also had to understand the physics. When a run failed, I had to work out whether the fault was in my code or in the materials science.
Co-founded
Plarick
A hollow, interlocking brick made from waste plastic, built for temporary housing in shelters, camps and schools.
- Team
- Aarav Jain, Aarav Jain and me
- Material
- HDPE, PET and PP with sand and additives
- Weight
- About 4 kg, 35% hollow
- Result
- First place, Imperial STEM-a-thon India; first place, Imperial FoNS International Science and Innovation Competition
I co-founded Plarick with two schoolmates for Imperial College London's Schools STEM-a-thon in India. The 2024 challenge asked teams to address one of the UN Sustainable Development Goals. We chose SDG 11 and set out to turn plastic waste into a brick that could replace clay.
The brick is 35% HDPE for strength, 10% PET to cut weight and water absorption and 15% PP for heat and UV resistance. The rest is 25% sand for binding and 15% calcium powder with Teflon and carbon black, which limit shrinking and improve fire resistance. The mix is heated and set in moulds.
In our tests the brick absorbed 0.7% of its weight in water, against 11.5% for a clay brick. We worked out a cost of about ₹37 per brick at 1,440 bricks a day.
How the brick changed
Ratios
We set the first mix from published research. The first brick was flat.
Additives
We changed the ratios and added limestone, Teflon and calcium.
Interlocking
We were not happy with a flat brick and made an interlocking one in a makeshift compression mould.
Hollow
The final brick is 35% hollow, which cut its weight from 7 kg to about 4 kg.
Competitions
December 2024, Delhi
Imperial's Schools STEM-a-thon India
Our team won first place at the national final in Delhi, after a pitch video and a business plan. The win gave us a place in the final of Imperial's Faculty of Natural Sciences competition in London.
March 2025
Letter of recommendation from Imperial
Clare Whelan of Imperial's Global Summer School wrote a letter of recommendation after the competition.
June 2025, London
Imperial FoNS competition
We presented Plarick at the Faculty of Natural Sciences International Science and Innovation Competition on Imperial's South Kensington campus, and won the 16 to 19 age group.
Team competition
Space Settlement Design Competition
Two years designing a space settlement with my school's team, first as Head of Structural Engineering and then as Team Captain.
- Year one
- Head of Structural Engineering
- Year two
- Team Captain
- Nationals
- Runners-up, 2024
- Company
- Grumbo Aerospace
In the competition, teams act as aerospace companies and design a settlement for thousands of people under a tight deadline. My school entered for the first time when I was in Grade 10, and I led the structural engineering team.
Our settlement, Aresam, has a main station called HighPort in areostationary orbit about 17,032 km above Mars, with room for 15,000 permanent residents. A smaller station, MarsPort, sits 200 km above the surface for up to 2,000 people on their way to Mars missions.
Structure
HighPort
HighPort has two tori stacked 50 m apart around a central cylinder. Torus A houses 10,000 people. Torus B houses 5,000 and holds the administrative areas, labs, farms and the control centre. Six spokes link each torus to the centre and carry power, water, air and people. The docking ports sit on top. The fusion reactor sits at the bottom, far from the living areas, in a section that can be detached if it fails. Xenon ion thrusters keep the station in orbit.
The tori have an inner radius of 50 m and reach 851.75 m from the centre. They turn once every 65 seconds to create gravity close to Earth's.
MarsPort uses the same design at a smaller scale. At 310 m and one turn every 50 seconds it gives 0.5g.
Rotation
The hardest structural problem was the link between the spinning tori and the central structure. The tori have to turn to make gravity, while the centre has to stay rigid enough to hold the docking ports and the reactor. We had four days to solve it.
In the final design, four rotation systems join the two tori. Each one is a motor and three gears held in a single housing. The gears drive the tori in opposite directions at the same angular speed, so their angular momentum cancels and the station does not twist as a whole.
Construction sequence
The settlement is built at a Lagrange point over about two months and then flown to Mars orbit. The links between the tori and the cylinder use a truss design, which carries large loads with little mass and spreads the stresses from rotation evenly.
1. Central cylinder 2. Spokes linking out from the cylinder 3. Tori assembled from the spokes 4. Fusion reactor attached 5. Motor and gear rotation systems 6. Rotation started, utilities installed
Materials
The hull is a layered shell. A 10 mm titanium alloy layer carries the structural load. A 75 mm layer of zirconium with aluminium oxide insulates against heat. A 0.2 mm carbon nanotube layer conducts electricity, a second 2 mm zirconium and aluminium oxide layer adds insulation, and a 2 mm titanium oxide coating blocks UV.
Homes inside HighPort are built from sintered regolith bricks on aluminium frames. The bricks are strong for their weight and shield against radiation and heat, and the aluminium carries the load-bearing walls.
November to December 2024
8th Indian National Space Settlement Design Competition
Grumbo Aerospace finished as runners-up at the national competition and qualified for the Asian Regional round. I returned the next year as Team Captain.
Co-founded
Project Tide-Y
A low-cost, solar-powered boat designed to clean drains and sewers so that workers no longer have to go into them.
- Team
- Aanyah, Arav, Ayan and me
- Role
- Co-founder and team lead
- Result
- Runner-up of 50+ teams
- Event
- SBS Model Shark Tank 2025
Blocked drains in Indian cities cause flooding and spread water-borne disease. They are still often cleared by hand. Manual scavengers climb into sewers with little protection, and many are hurt or killed by the gases inside.
Tide-Y is our answer. It is a small boat that moves along a drain and collects floating waste, so the cleaning can be done by a machine. It runs on solar power to keep running costs low.
Design
Our model
The hull has a camera at the bow and a motor at the stern. Two semicircular mesh nets on arms at the back collect waste as the boat moves. The hull is 1 ft 4 in wide at the top, 8 in at the base and 8 in deep. The flat roof is 2 ft 11 in long and 1 ft wide.
Getting it used
Working with the NAMASTE scheme
The government's NAMASTE scheme, the National Action for Mechanised Sanitation Ecosystem, aims to end manual scavenging and move sewer cleaning to machines. We planned Tide-Y around it in three levels.
Approved vendor
Tide-Y registers as an approved vendor under the scheme.
Public purchase
State and city authorities buy the boats using the scheme's capital subsidy.
Into workers' hands
The boats reach sanitation workers directly or through urban local bodies, at a subsidised cost.
The project works towards SDG 3, Good Health and Well-being, SDG 6, Clean Water and Sanitation, and SDG 10, Reduced Inequalities.
SBS Model Shark Tank 2025
Runner-up of 50+ teams
We pitched Tide-Y at the SBS Model Shark Tank 2025 at Step By Step School and finished as runner-up out of more than 50 teams.