PurvaJagtap
Full Stack Software Engineer in Charlotte, NC. I ship production backend services, distributed data pipelines, and LLM-powered applications end to end.
Drag the coin into the slot for a surprise, tap a snack, or type A3
About
Read the label. Hover a row.
Full stack software engineer with 3+ years shipping production backend services, distributed data pipelines, and LLM-powered applications across Python, Java, React/TypeScript, PostgreSQL, and AWS/Azure.
Most recently I built AI governance and GenAI cost tooling at Auditrol. I'm finishing my M.S. in Computer Science at UNC Charlotte (Dec 2026).
- Backend (Python/Java)0%
- Data pipelines0%
- Cloud (AWS/Azure)0%
- GenAI & LLMs0%
- Frontend (React/TS)0%
Ingredients: Python, Java, FastAPI, Spring Boot, React, PostgreSQL, Airflow, AWS, Azure, LLMs, curiosity.
May contain traces of coffee.
*Percent values are self-reported and deeply unscientific.
Experience
Your receipt is printing. Tap a line item for details.
Every role, itemized.
From Spring Boot microservices at Amdocs to AI governance at Auditrol. 3+ years, no hidden fees.
- › Shipped the AI Governance lineage graph across FastAPI and React, so auditors can trace any finding to the model and prompt that produced it.
- › Built GenAI Usage & Cost and NIST control-coverage backends joining live AWS Cost Explorer spend with per-model and per-use-case telemetry, with hermetic CI tests.
- › Fixed contradictory control statuses across two screens by sourcing verdicts from the ml-service system of record over server-to-server HTTP.
- › Cut peak memory on risk exports to one batch by streaming rows through an async generator with batched verdict lookups.
- › Landed daily SAS Viya data quality snapshots from S3 into PostgreSQL via a four-stage Airflow pipeline, surfacing 453 breaches across 26 attributes with z-score and PSI drift detection.
THANK YOU FOR VISITING · COME BACK SOON
Projects
Collectible cards. Tilt them, then tap to flip.
PharmaLLM Medicine Prescriber Chatbot
PharmaLLM Medicine Prescriber Chatbot
Fine-tuned TinyLlama-1.1B with LoRA and 4-bit quantization on 11,000 samples, reproducing published Springer research. Multimodal stack: Whisper speech-to-text, gTTS, Flask inference API, React frontend.
CareLess: AI Patient Advocate
CareLess: AI Patient Advocate
Agentic AI assistant for healthcare cost transparency, combining voice, blockchain payments, and financial APIs. Built in 24 hours. 3rd Place overall and Best Use of Solana at Pearl Hacks 2026.
Smart Attendance Monitoring System
Smart Attendance Monitoring System
Computer vision attendance system with 90% face detection accuracy, deployed across 10+ institutes serving 100+ professors. 1st Prize, TECHCULT 2022.
Skills
- 01
Languages
PythonJavaTypeScriptJavaScriptSQLBash
06 - 02
Backend
FastAPISpring BootREST APIsMicroservicesWebhooksRedispytest
07 - 03
Frontend
ReactStreamlitFlaskHTML/CSS
04 - 04
Data & Cloud
PostgreSQLMySQLAirflowAWSAzureDockerKubernetesCI/CD
08 - 05
AI / ML
LLMsRAGAgentic workflowsLangChainPGVectorPyTorchHugging FaceLoRA
08
Education & Awards
Tear a ticket. Ring a medal.
M.S. Computer Science
University of North Carolina at Charlotte
B.E. (Honors) Computer Engineering, Data Science
Savitribai Phule Pune University
AT&T Radiant Recognition
For the GenAI Deployment Tracker used by 5,000+ engineers.
3rd Place + Best Use of Solana
Pearl Hacks 2026, for CareLess: AI Patient Advocate.
1st Prize, TECHCULT 2022
National level project competition, for Smart Attendance.
Contact
Put a message in the can and send it off.
Got a problem worth building for? Let's talk.