Open to full-time SWE & AI/ML roles · from Feb 2027

I build software that makes AI actually useful.

I'm Aniket, an MS Computer Science student specializing in AI/ML at the University at Buffalo. I build retrieval-augmented LLM apps and full-stack products, and I stress-test frontier coding agents at Handshake AI.

See my work Résumé
aniket@buffalo: ~ try typing
RAG grounding accuracy in DocuMind
92%
critical limitations found in frontier coding agents
14
financial transactions a day through my .NET APIs
10k+
registered users on a platform I led
4,000+
01 · Selected work

Things I've built

Three projects, each with something to click on: a replay of a RAG experiment, a narrated product walkthrough, and a smart-contract flow you can run.

AI · Retrieval-augmented generation Live

DocuMind

A multimodal RAG assistant for technical documents. Instead of pasting a whole PDF into a chatbot every time, it indexes the document once, retrieves only the evidence each question needs, and has an LLM answer from that evidence.

Try it live GitHub Runs on a free tier. If the backend is asleep, the first request takes about 30 seconds.
  • 3 / 3answers matched a full-document chatbot
  • 92%contextual grounding accuracy
  • −50 msinference latency with caching
  • −20%API cost
  • FastAPI
  • React
  • FAISS + BM25
  • sentence-transformers
  • OCR
  • Docker
  • OpenAI / Groq (BYOK)
Full-stack · MEAN UB CSE 611 · Demo Day

Safety Knights

A careers and community platform for health & safety professionals, with a job board, saved jobs, applications, a résumé vault and a full admin console. I led the full-stack build. Below is my narrated Demo Day walkthrough.

Safety Knights · Demo Day walkthrough Highlights · muted
ChaptersClick to jump in

Test-account names and emails are blurred for privacy.

What I built

  • Led full-stack development on the MEAN stack, including an admin dashboard for tracking content and users across 500+ listings and 4,000+ users.
  • Modular Angular components with state-driven UI for real-time job creation, search and tracking.
  • A normalized MongoDB schema with indexed queries that cut data-retrieval latency by 40%.
  • Middleware validation, role-based access control and protected API routes.
  • 4,000+registered users
  • 500+active job listings
  • −40%data-retrieval latency
  • MongoDB
  • Express
  • Angular
  • Node.js
  • REST
  • RBAC
Blockchain · Full-stack

Hybrid land registry

Property records on Ethereum, checked by the authorities who already hold the paperwork. I designed and built the platform end to end. Smart contracts make ownership tamper-proof, and a government validation layer keeps it legally meaningful.

An illustration of the flow. Names, balances and block numbers are examples; gas is paid in ETH (Ξ) because the platform runs on Ethereum.

  • 60%faster property verification
  • −30%gas fees after contract optimization
  • <15 stransaction finality
  • Ethereum
  • Smart contracts
  • Ethers.js
  • MetaMask
  • REST APIs
02 · Experience

Where I've worked

  1. AI Trainer · Coding Agent Evaluation

    Handshake AI · Remote

    May 2026 — Present Now
    • Designed 30+ adversarial command-line evaluations, complex multi-step tasks in WSL and tmux that surfaced 14 critical limitations in frontier AI coding agents within two months.
    • Wrote domain-specific prompts for the Panda3D and Solar3D engines to generate and refine working game models, reaching an 80% learnability rate.
    • Review and revise hundreds of LLM outputs into high-quality annotated data that improves the models' autonomous reasoning and code execution.
    • LLM evaluation
    • Prompt engineering
    • WSL
    • tmux
    • Panda3D

  2. Web Developer Intern

    SmowCode · Mumbai, India

    Dec 2023 — Feb 2024
    • Led the migration of the public homepage from WordPress to React + Next.js, redesigning the UI and cutting page load time by 25%.
    • Built SEO-optimized, responsive components with server-side rendering and dynamic metadata, lifting organic search traffic by 15%.
    • Refined the homepage with designers and founders based on usability feedback, reducing bounce rate by 20%.
    • React
    • Next.js
    • SSR
    • SEO

  3. Backend API Developer Intern

    Workflowed Solutions Pvt. Ltd. · Mumbai, India

    May 2023 — Jul 2023
    • Built secure RESTful APIs in .NET processing 10,000+ sensitive financial transactions daily, speeding up internal data processing by 30%.
    • Designed new API modules for specific business use cases while keeping backward compatibility.
    • Added authentication and authorization to secure access to financial data, then tested and documented every endpoint.
    • C# / .NET
    • REST
    • Auth
    • SQL
03 · Toolkit

What I work with

Machine learning & AI

  • PyTorch
  • TensorFlow
  • Hugging Face Transformers
  • RAG
  • FAISS
  • BM25
  • sentence-transformers
  • LLM fine-tuning
  • LLM evaluation
  • CNNs
  • NLP
  • scikit-learn
  • XGBoost
  • Optuna

Languages

  • Python
  • TypeScript
  • JavaScript
  • C# / .NET
  • C / C++
  • SQL
  • R
  • Bash

Web & backend

  • React
  • Next.js
  • Angular
  • Node.js
  • Express
  • FastAPI
  • Flask
  • MongoDB
  • REST APIs

Cloud & tools

  • AWS
  • GCP
  • Docker
  • Git / GitHub
  • Vercel
  • Render
  • WSL
  • tmux

Data & Web3

  • pandas
  • NumPy
  • Matplotlib
  • Ethereum
  • Ethers.js
  • MetaMask
04 · About

A little about me

I like working at the seam between machine learning and real software, where a model has to sit behind an API, answer quickly, cost little, and be right often enough that people trust it.

Before grad school I shipped production code in Mumbai: .NET APIs moving sensitive financial data at Workflowed Solutions, and a Next.js rebuild of SmowCode's public site. At UB I've gone deep on deep learning, pattern recognition and computer vision, and built DocuMind and Safety Knights.

These days I write adversarial terminal tasks that expose where frontier coding agents break. It has taught me a lot about how LLMs reason, and how they fail.

Based in
Buffalo, NY
Studying
MS CSE (AI/ML), UB
Graduating
December 2026
Available
February 2027
Open to
SWE · AI/ML · Data
Relocation
Anywhere in the US
Aug 2025 — Dec 2026

MS, Computer Science & Engineering

University at Buffalo (SUNY) · AI/ML specialization

  • Algorithms
  • Machine Learning
  • Deep Learning
  • Pattern Recognition
  • Computer Vision
Aug 2021 — May 2025

BE, Computer Science

Rajiv Gandhi Institute of Technology, University of Mumbai

Certifications

  • Associate AI EngineerDataCamp
  • Advanced Learning AlgorithmsDeepLearning.AI & Stanford · Apr 2025
    Verify
  • Supervised Machine Learning: Regression and ClassificationDeepLearning.AI & Stanford · Jan 2025
    Verify
05 · Contact

Let's build something useful.

I'm looking for full-time software and AI/ML engineering roles starting February 2027, anywhere in the US. Email is the fastest way to reach me.

Email me LinkedIn GitHub