AI/ML ENGINEER · DATA SCIENTIST — PITTSBURGH, PA

I build AI systems that hold up outside the demo.

2+ years shipping computer vision, LLM/RAG, and cloud analytics — currently finishing a master's at Carnegie Mellon (MISM-BIDA, Dec 2026).

ASK MY AIllama-3.3-70b · ⌘K

What do you want to know about my work?

LATEST

AUG 2026Built this site. The chat above is my real LLM work running live — tool calls, guardrails, and a fallback for when the model is down.
AUG 2026Wrapping up at Kinertia: turned 3.3 years of flight data into a daily analytics product that leadership greenlit for public launch — and cut the query bill from $6 to 2¢ per run.
JUL 2026Brigade Kitchen — a game where one human and three AI agents run a kitchen together, built for CMU's Dynamic Decision Making Lab to study how humans and AI team up.
SPRING 2026Trained a transformer that tells which kind of image you're looking at from EEG brainwaves alone — 2.5× better than chance.

SELECTED WORK

01Beyond Vector Search — four ways to search a 200-page SEC filing, tested head-to-head

github ↗

02EEG Visual Decoding — reads which image you're looking at straight from EEG brainwaves — 2.5× over chance

github ↗

03Brigade Kitchen — one human and three AI agents run a kitchen together — CMU's testbed for human-AI teamwork

04Adaptive RAG Router — a Q&A service that decides on its own whether to search its documents or the web

EXPERIENCE

Kinertia (Wing Jockey)

Data Analytics Intern

Jun – Aug 2026 · Pittsburgh, PA

  • Built a production Airflow DAG on GCP BigQuery aggregating 32M+ ADS-B flight legs (2.7 TB, 3.3 years) into daily aircraft- and airport-level statistics; cut per-run scan cost from ~$6 to ~$0.02 (300×) via partition and columnar pruning.
  • LLM-powered classification pipeline labeling all 847 aircraft type codes into a dimension table; validated 96.8% coverage over 2.6M flights against an independent heuristic.
  • Prototyped a public flight-analytics site (airport dashboards, aircraft 'baseball cards', national heat map of 8,756 airports); demoed to leadership and greenlit for publication.

CMU Dynamic Decision Making Lab

AI Engineer Research Assistant

Jun – Aug 2026 · Pittsburgh, PA

  • Designed and built Brigade Kitchen, a parameterized 4-agent cooperative environment for human-AI teaming experiments; design approved by lab PI.
  • Implemented rule-based agents (BFS + priority FSMs) and per-step JSONL telemetry for behavioral analysis; deterministic replay; 3 experiment layouts.

Sampark Softwares

AI/ML Engineer

Jun 2024 – Jun 2025 · India

  • Owned end-to-end rollout of a Dockerized, high-concurrency Flask inference API for YOLOv11, scaling to 1,500+ stores.
  • Automated a daily web-scraping ETL pipeline for regional competitor data; shipped self-serve Power BI benchmarking dashboards contributing to a 4% national operational revenue increase.

Sampark Softwares

AI/ML Intern

Jan – May 2024 · India

  • Reduced LLM query latency 30% (15s to 10s) during an OpenAI-to-Azure migration via caching and request-flow optimization.
  • Orchestrated parallel Selenium data collection across 1,200 RTO regions; centralized MySQL analytics improved regional sales targeting 5%.

SKILLS

PROGRAMMING

Python · SQL · R · Java

ML / LLM

PyTorch · scikit-learn · NLP · SentenceTransformers · LangChain · LangGraph · LLM-based labeling · evaluation/metrics

MLOPS / DEPLOYMENT

FastAPI · Flask · Docker · Airflow · model serving · monitoring/logging · Playwright

DATA

BigQuery · ETL · Spark · Kafka · MySQL · partition/cost optimization

CLOUD / BI

GCP · AWS · Azure · Power BI

VECTOR / RAG

ChromaDB · AstraDB (Cassandra) · Neo4j · RAG architectures

CONTACT

emailgithublinkedinresume