JEE Advanced selection into IIT Kharagpur
Bangalore, India · Learning across domains
I learn hard systems.
Then I make them useful.
I’m Kumar Nilay, a quantitative technical founder and product builder working across AI, financial systems, security, privacy, and the messy space between an idea and a working product.
01
Chemistry to computation
Science
I cleared JEE Advanced to enter IIT Kharagpur, studied chemistry, added computer science, and learned to move between scientific and computational models.
recognized projects and programs
aggregate project awards
smart-contract security audits
One career, several hard systems
I have kept learning new domains without changing how I learn.
I enter an unfamiliar system, find the mechanism that matters, build a working proof, and stay close enough to the outcome to learn what the model missed.
Selection pattern · with evidence
I keep reaching the narrow end of new fields.
Not every outcome is literally top 1%. The repeatable signal is that different selection systems keep producing the same result: admission, medals, fellowships, product prizes, and a startup final.
BioVenomSDK
I built it solo and won the Venom Hackathon Tools & Infrastructure track.
JEE Advanced
Rank 6,671 in the national IIT admission pipeline, from about 1.3 million candidates.
ETHIndia Fellowship
Selected for the Gwei senior track from the top 0.5% of applicants.
Inter IIT Tech Meet
Our autonomous cleaning robot won the competition across participating IITs.
Solana Breakpoint Fellow
Selected as one of 50 young engineers worldwide for the under-25 Global Fellows program.
Startup World Cup
AlphaEngine reached the final at Devconnect after placing in the top 6 of 119 at qualifiers.
01
Learn · 2015-2020
I started in chemistry, then kept crossing boundaries.
At IIT Kharagpur, my journey moved from chemistry to machine learning, then computer vision, autonomous robotics, and FPGA systems.
A
IIT Kharagpur · Autonomous robotics
I learned by making machines confront the physical world.
2018Cleaning robotInter IIT Tech Meet gold medal2019Fruit harvesterStudent project, published at ICCAR 2020
EntryTop ~0.5%Cleared JEE Advanced for IIT Kharagpur
I led autonomous robotics work across fruit harvesting, cleaning, perception, navigation, stereo vision, FPGA inference, and team execution.
B
Vector Institute to IIT Kharagpur thesis · 2019-2020
Generative ML for molecular design
My Vector Institute internship became the core of my master’s thesis. I worked on a joint semi-supervised VAE and property-prediction network for molecular generation.
78.7% molecule reconstruction92.6% property prediction
View the complete thesis 
Figure 1 from the thesis: joint molecular generation and property prediction architecture.
02
Professional work · 2020-2025
I learned to own systems that other people depended on.
My full-time work moved from institutional risk, to protocol rules, to the contracts and deployment infrastructure supporting capital. Select each role to see the system I was responsible for.
Full-time work / Jul 2020 - Mar 2022
I learned how models behave inside an institution.
I developed the core MXP framework used for market-risk analysis, contributed to the Python 3 migration and FRTB work, and later moved into credit-risk development.
Core risk framework across the businessPython 3 migrationFRTBMXP / core market-risk framework
03
Secure · 2022-2025
I learned to think like the person trying to break the system.
Security changed how I build. Every interface carries assumptions; every recovery path creates a new attack surface; every privacy promise needs a mechanism behind it.
Oak Security
Adversarial review
I contributed to four part-time audits spanning EVM and Solana systems.
Wallet recovery
Security that remains usable
I explored biometrics, MPC, ZK, randomness, and recovery flows across several working prototypes.
Private computation
Protect the action, not only the key
I built with FHE and privacy infrastructure to explore confidential market interactions.
04
Build · Selected work
I use prototypes to ask sharper questions.
Explore the projects by domain. Each preview leads to public evidence instead of a decorative case-study page.
Selected buildFinance
Project / Finance
AlphaEngine
A quantitative strategy evaluation and confidential-execution platform with a live testnet beta.
Final 12 at the Startup World Cup on Devconnect’s main stage; official Fhenix case study.
05
Found · 2025-now
I now own the whole question, not only the implementation.
AlphaEngine became a test of whether I could connect quantitative research, product architecture, privacy infrastructure, positioning, partnerships, and user evidence into one coherent operating loop.
Founder lens / Product
Turn strategy research into something people can try.
I shaped a live testnet beta and a catalogue of about 54 strategy types around simulation, evaluation, and comparison.
Recognition, with receipts
A repeatable builder signal, not a trophy wall.
Thirteen project, fellowship, grant, and developer-program outcomes; more than $76K in aggregate project awards. Selected highlights:
Continue the conversation
If this work overlaps with a problem you care about, let’s compare notes.
I share this site as a record of how I learn, build, and make decisions across hard systems. If an idea, project, or technical question here connects with your work, feel free to reach out.