About
Computer Engineering student at MES Wadia College of Engineering with hands-on experience in AI engineering, LLM evaluation, and cybersecurity training. Built production-oriented ML systems and agentic AI projects using modern Python ML and cloud tooling.
Experience
-
AI Engineer InternQ3 Learners LLC · USA (Remote)Feb 2026 – May 2026
Owned the end-to-end data pipeline and model integration layer for the company's AI-education product as sole ML engineer on a 3-person team shipping to production. Shortened prompt-iteration cycles and improved output consistency across LLM version upgrades by designing A/B prompt-evaluation protocols with quantitative model-version comparison. Researched LLM alignment for adaptive tutoring, comparing in-context learning vs. fine-tuning to guide the approach for educational domain specialization.
-
Cybersecurity InternAICTE-Edunet / IBM SkillsBuildJan 2025 – Feb 2025
Built foundational skills in network security, threat analysis, and IBM security tooling through a structured cybersecurity program, completing all program assessments.
Education
-
2027
B.E. Computer EngineeringMES Wadia College of EngineeringComputer Engineering · Aug 2023 – May 2027
Skills
Tools / apps / platforms
Projects
-
Agentic AI System for Chronic Kidney Disease (CKD) ManagementIBM Granite 3.2 · Watsonx.ai · LangFlow · AstraDB
Winner at VOIS Innovation Marathon 2.0 and secured a hospital pilot at Vignahartha Hospital. Built a 6-agent clinical pipeline achieving 91% sensitivity / 87% specificity on ESRD risk stratification and 89–94% retrieval accuracy on RAG stage-detection over 10,000+ nephrology papers. Developed a multilingual virtual nephrology agent supporting 15+ Indian languages.
-
SOC Simulator — RL Environment for CybersecurityFastAPI · PyTorch · TRL · Unsloth · GRPO · MITRE ATT&CK
Placed Top 15 globally out of ~31,000+ teams at the Meta × PyTorch OpenEnv Hackathon Finals, scoring 91/100 at onsite finals. Built an OpenEnv-compliant RL environment spanning 3 MITRE ATT&CK scenarios, with a Schema Drift Engine that randomizes log field names per episode to prevent policy overfitting. Achieved a +0.551 reward uplift over baseline on a hard APT task with the GRPO blue agent; fully reproducible via open repo with W&B training logs.
🏆 Achievements & awards
-
Finalist — Meta × PyTorch OpenEnv Hackathon · 2026
Top 15 out of ~31,000+ teams, scored 91/100, onsite finals at Scaler Bengaluru.
-
Winner — VOIS Innovation Marathon 2.0 · 2026
Agentic AI on IBM Granite & Watsonx (₹2L prize).
⚽ Extracurricular activities
-
Research Paper
Securing YOLOv8 for Pothole Detection: A Hybrid Defense Framework Against Adversarial Attacks