Applied Machine Learning Engineer (MTS-1 · AI R&D) at Mavenir, Bengaluru — working on applied AI research, large-scale experimentation, and production-grade intelligent systems.
Specialized in LLMs, Agentic AI, Reinforcement Learning, Fine-Tuning, NLP, and Computer Vision, with strong interests in optimization, statistical learning, and mathematical foundations of AI. Passionate about first-principles thinking, model behavior analysis, and building scalable AI systems grounded in research and real-world impact.
Apr 2026 – Present | Bengaluru, Karnataka, India · Hybrid
Working in the Data & AI R&D team on applied AI research, LLM fine-tuning, and production-grade intelligent systems across telecom-scale infrastructure.
Focus Areas: Fine-Tuning · Large Language Models · Agentic AI · Reinforcement Learning · NLP · Computer Vision
Jul 2025 – Apr 2026 | Noida, Uttar Pradesh, India · On-site
- Built an AI-powered boiler efficiency recommendation system using Gradient Boosting models with a Differential Evolution optimization engine — supporting what-if simulations, target-efficiency optimization, and operational safety bounds.
- Deployed a production-grade LLM-powered industrial IoT analytics platform using FastAPI, LangGraph agentic workflows, GPT-4, and vector databases — enabling natural-language querying over 1000+ real-time time-series sensors.
- Implemented advanced time-series forecasting & anomaly detection pipelines using Chronos, TimesFM, LSTM Autoencoder, and TSPulse across 10 critical industrial KPIs.
- Designed a scalable multi-tenant microservices architecture with Docker, JWT, pgvector, MongoDB conversational memory, WebSockets, and distributed caching.
Apr 2025 – Jun 2025 | Bengaluru, Karnataka, India · On-site
- Optimized YOLO-based object detection pipelines for industrial defect detection — achieving ~35% faster inference, ~28% accuracy improvement, and ~45% reduction in false positives.
- Conducted R&D on complex defect categories (rust, tilt, surface/structural anomalies) with custom augmentation strategies and class imbalance handling.
Sep 2024 – Mar 2025 | Lucknow, Uttar Pradesh, India · On-site
- Developed an end-to-end AI thunderstorm prediction system for Uttar Pradesh using a BiLSTM model on 24-hour meteorological sequences, achieving ~69% accuracy and 0.74 AUC — outperforming traditional stability-index methods.
- Built a production-ready Flask web app with real-time geospatial heatmaps, lightning tracking, and 12-hour forecasts for early-warning systems.
MSc in Artificial Intelligence and Machine Learning Indian Institute of Information Technology Lucknow · Aug 2023 – Jun 2025 · CGPA: 8.20
BSc (Hons) in Statistics Maharaja Bir Bikram University · Dec 2020 – Jul 2023 · CGPA: 7.20
- 🥇 IIT JAM 2023 — All India Rank 218 (Top 7%) in Statistics
- 💻 CodeSmash 2.0 — Ranked 171st out of 3,000+ participants
- 📊 Kaggle — 0.81 score in PII Detection Challenge
- 🛰️ ISRO Certified — AI/ML for Geodata Analysis
- 🏅 IIIT Lucknow — Gold Medalist · Chess (INFINITO)
-FF9900?style=for-the-badge){kind=icon}
-4285F4?style=for-the-badge){kind=icon}
-052FAD?style=for-the-badge){kind=icon}
Working at the intersection of Applied Research × Production AI. Always open to discussing LLM fine-tuning, agentic systems, RL, and large-scale AI engineering.