Meta via HCL Technologies
ML Infrastructure Engineer
Built Sherlock, an AI assistant that resolves ML-infrastructure questions on its own, plus the monitoring and tooling that keep model training and deployment running.
Sambhav
Shrestha
Software engineer working on the infrastructure behind machine learning. Recently at Meta; now at Stony Brook for a master's in computer science.
Work
Tarifica
Software & Data Engineer
Built the pipelines that validate and route telecom pricing data into analytics, and refactored a legacy codebase with logging, observability, and tests.
Amazon
Software Development Engineer
Operated production microservices and led a high-traffic service's migration to AWS.
Microsoft Research DS3
Data Science Fellow
Extended the Financial Times police complaints study with statistical models in R.
Projects
Sherlock
An AI assistant built at Meta that answers engineers' ML-infrastructure questions, trained on internal forum data.
LLaMA · RAG · PyTorch
Argument Quality Ranking
A model that ranks how well-made a written argument is.
Python · NLP
GitHub
Raft / Paxos Consensus Simulator
Watch distributed systems agree, disagree, and recover from failures.
Go
GitHub
Pipelined MIPS Processor
A 32-bit five-stage CPU with caches, built from scratch.
Verilog
GitHub
Frequency Prior for Image Generation
Replacing attention with a frequency-based prior in image generation models.
Python · PyTorch
GitHub
Stock Price Prediction
An LSTM that forecasts stock prices from historical data.
Python
GitHub
Police Complaints Analysis
Race and gender patterns in police complaints across NYC, Chicago, and Philadelphia, at Microsoft Research.
R
GitHub
Credit Card Fraud Detection
Catching rare fraudulent transactions in an imbalanced dataset with XGBoost and SMOTE.
R
GitHub
Education
Stony Brook University
M.S. Computer Science
St. Joseph's College New York
B.S. Computer Science & Mathematics
GPA 3.93 · President's Scholar