💡 I build AI systems that don't just demo well — they hold up at scale. From architecting multi-agent pipelines on a $10M NSF-funded AI program to shipping production RAG and inference services processing 120M+ tokens/day at 99.95% uptime, I turn cutting-edge AI research into infrastructure that survives real traffic, real cost pressure, and real deadlines. I've led teams, mentored 50+ engineers, and consistently cut latency, cost, and incident rates by double digits everywhere I've worked — across GCP, AWS, and everything in between.
- 🤖 AI Engineer @ Handshake AI, working on cloud-scale ML infrastructure and visualization systems
- 🎓 Formerly built multi-agent LangGraph pipelines & production RAG systems at Indiana University, part of a $20M NSF-funded AI program
- 👩🏫 Mentored 50+ graduate students in LLMs, transformers & fine-tuning as a Graduate AI Instructor
- 🏗️ Started out as a Data Engineer at Cognizant, building real-time pipelines for General Mills & Bayer
- 📜 AWS Certified AI Engineer · Neo4j Certified Professional
- 🌱 Currently exploring agentic tool-use, vector search at scale, and cost-efficient LLM inference
Impact at a glance
| 🚀 65% | 📊 120M | 🧵 15K+ | 💾 15TB+ | ⏱️ 99.95% |
|---|---|---|---|---|
| deployment latency cut | tokens/day processed | concurrent requests served | data processed in Spark pipelines | RAG system uptime |
Languages & Databases
AI / ML
Data Engineering
Cloud & DevOps
|
🩺 Autonomous Clinical Triage Agent Network Agentic RAG services parsing symptoms and assigning case priority — improved triage accuracy 41%, cut routing delays 38% via Kafka Streams + PostgreSQL decision agents.
|
🌦️ WeatherMind — AI Weather Intelligence App Full-stack AI weather assistant with agentic tool use for real-time city comparisons and recommendations, streaming multi-turn chat, deployed on Vercel.
|
- ⚡ Cut deployment latency 65% by migrating log processing to client-side execution across 11 hardware profiles
- 📊 Built RAG systems over 500K+ records, processing 120M tokens/day at 99.95% uptime
- 🚀 Reduced p99 latency 40% under 15K+ concurrent requests with FastAPI + Redis
- 💾 Ran distributed PySpark pipelines over 15TB+ datasets, cutting runtimes 45%
- 📈 Reduced MTTR 30% with a Prometheus + Grafana observability stack
- 🎓 Indiana University Bloomington — M.S. Data Science (Aug 2023 – May 2025)
- 🎓 Sikkim Manipal University — B.Tech Computer Science (Aug 2017 – Jun 2021)
📫 Reach me at ayantika.n18@gmail.com