Challenge Orthodoxy
Innovator, Builder
Welcome to my website. It is a decentralized static React application deployed on the InterPlanetary File System (IPFS). It is also deployed through in the Ethereum Name Service (down for maintenance). You can learn more about me in the sections below. You can navigate the site and find my Github, ORCID, socials via the side panel.
There is nothing impossible for those who will try.
Engineer, Scientist, Entrepreneur, Inventor
I am an AI full-stack developer and researcher. I'm a TechBio accelerationist developing ML methods and software systems in the biomedical and life sciences. My mission is to drive humanity forward by increasing our population, extending lifespans tenfold, and expanding human consciousness across the universe. I firmly believe that machine learning will accelerate scientific discovery at a pace previously seen only in software, unlocking the pinnacle of human progress.
Harvard Medical School
( 2024 — 2025 )Research Fellow
Built PyTDC, A multimodal machine learning training, evaluation, and inference platform for biomedical foundation models. Built llm agents and foundation models for drug discovery and biomedical applications.
Cruise Automation LLC and various startups
( 2021 — 2022 )Senior Machine Learning Software Engineer
Developed uncertainty quantification pipeline for computer vision models and machine learning infrastructure for data labeling platform. Decided to leave to pursue entrepreneurship and more generally a career break.
Pinterest Inc
( 2018 - 2021 )Software Engineer II, Data and Machine Learning
I worked on a variety of exciting projects, starting with large scale distributed systems for data analyics, the online a/b experimentation platform, a gpu frameowork for streamlining AI inferencing and deployment, and optimizations on the Pinterest Ads ranking model.
Massachusetts Institute of Technology
B.S. Computer Science
Minor in Management, Concentration in Economics
An AI full-stack developer with 4+ years of experience in building AI applications and platforms. Additional experience in developing, training, benchmarking, and deploying large-scale models (LLMs, ranking models, etc.). Have also built LLM agent systems. Most recent experience has been in biomedical domain, but I have worked on a variety of domains.
You define your own life. Don’t let other people write your script.
Featured articles and projects
April 26, 2019
My article featured in the Pinterest Engineering Blog
OpenAI, Javascript, Full-stack development
EEGPT - Brain Waves LLM. Non-invasive BCI
OpenAI, Javascript, Full-stack development
EEGPT - Brain Waves LLM. Non-invasive BCI
Generative AI meets at-home neurodegenerative disease monitoring and care. A non-invasive BCI app integrated with Neurosity EEG for monitoring and alerting of vital brain activity from the comfort of their homes.
React.js, Javascript, Python, Full-stack development
NFT Search Engine and Masonry
React.js, Javascript, Python, Full-stack development
NFT Search Engine and Masonry
I've found the webpage design for most nft marketplaces to be inadequate. They do a good job of showing the popularity and trading activity of NFTs. However, I don't believe they showcase the content of digital artwork well. So, I decided to build a search engine and masonry layout for NFTs. I hope it can help you find an NFT you like.
entrepreneurship, web3
Apliko: Web3 startup
entrepreneurship, web3
Apliko
I worked on launching a web3 startup and built a webapp with some thousand users. I wrote a whitepaper for a token and gave an interview about the concept here: https://nftcalendar.io/news/apliko-emblem-interview/.
LLM Agents, Computational Biology, BMI
Therapeutic LLM
LLM Agents, Computational Biology, BMI
Therapeutic LLM
AGI House Hackathon: Therapeutic LLM using function calling for disease association and single-cell analysis. Code: https://github.com/amva13/therapeutics-llm.
LLM Agents, Medical Informatics, RAG
Pediatrics LLM
LLM Agents, Medical Informatics, RAG
Pediatrics LLM
*Awaiting peer review*: Monitoring a child’s growth is vital for early detection of disorders. This study explores integrating a large language model (LLM) agent into the SMART on FHIR Growth Chart App to assist pediatricians in identifying growth abnormalities. Using a User-Centered Design (UCD) approach, we gathered pediatrician feedback to refine an AI tab analyzing synthetic patient data. The system was implemented using the OpenAI Assistants API with Retrieval Augmented Generation (RAG) and tested for usability and functionality with three pediatricians in an evaluation of model responses for growth abnormality detection, patient history analysis, recommended specialist referral, differential diagnosis and executive summary. Results showed that the agent can achieve high levels of usability when integrated into a clinical setting. However, while the agent accurately analyzed three of five synthetic patients, its responses to differential diagnoses and specialist referrals were insufficient. This proof of concept highlights the potential of AI tools in pediatrics but also underscores the need for improved accuracy in future developments. We have open-sourced the agentic app.
hedge funds, trading, Python, big data
Cryptocurrency ticker data infrastructure
hedge funds, trading, Python, big data
Cryptocurrency ticker data infrastructure
Developed system design and prototype implementation of data infrastructure serving quantitative research for a cryptocurrency hedge fund client.
data science, python, data engineering, quantitative research
Geometric Compactness in Gerrymandering: Research Assistantship MIT CSAIL
data science, python, data engineering, quantitative research
Geometric Compactness in Gerrymandering: Research Assistantship MIT CSAIL
Developed a high-performance geometric data processing library for scientific computing over GIS data (shapefiles) in Python.
Researched and implemented high performance geometric data processing algorithms for political redistricting. Scaled polygon adjacency algorithms for big data geospatial analysis.