PROFILE
• Senior AI Systems & ML Research Engineer specializing in high-concurrency production LLM architectures, Agentic workflows, and distributed inference.
• Deep expertise in bridging state-of-the-art Generative AI research with enterprise-scale serving infrastructure.
• Proven track record of designing autonomous multi-agent state machines, hybrid retrieval systems, and post-training alignment pipelines.
SKILLS SET
• Generative AI & LLMs: Advanced RAG, Post-training (SFT, RLHF/DPO), PEFT (LoRA/QLoRA), LLM Evaluation (LLM-as-a-Judge, HITL).
• Agentic Architectures: Autonomous AI Agents, Tool-Calling, LangGraph State Machines, Model Context Protocol (MCP), ReAct Planning.
• ML Systems & Stack: PyTorch, vLLM, DeepSpeed, Triton, HuggingFace, Ray, FAISS, Docker, Kubernetes.
• Infra & Analytics: Python, SQL, Distributed Computing, Large-Scale Feature Engineering, A/B Testing & CUPED.
WORK EXPERIENCE
• TalentMe AI Labs & Systems | Senior AI Systems Architect
◦ Architected TalentMe Autonomous Multi-Agent Engine orchestrating domain tools via Model Context Protocol (MCP), maintaining hierarchical state across sessions.
◦ Deployed high-throughput multi-tenant vLLM inference clusters for 70B+ open models, cutting P99 latency by 45% via PagedAttention and FP8 quantization.
◦ Engineered custom preference optimization (DPO) and reasoning RL pipelines, personalizing interactive interview simulations and pedagogical paths.
◦ Established automated end-to-end evaluation harness measuring generation fidelity, hallucination rates, and multi-turn state coherence.
• Enterprise Cloud & AI Platform | Machine Learning Engineer
◦ Developed hybrid semantic retrieval engine combining dense vector search (VectorDB) and BM25 with cross-encoder reranking, boosting retrieval precision by 34%.
◦ Implemented real-time feature drift and telemetry monitoring pipeline, reducing anomaly detection latency from weeks to minutes.
◦ Optimized distributed training and model serving infrastructure, cutting GPU compute overhead by 35% on multi-node clusters.
RESEARCH & PROJECTS
• TalentMe Tech Vault & Interactive Career Transition Engine
◦ Designed full-stack AI knowledge graph architecture linking 450+ official atomic concepts, interactive mindmaps, and role roadmaps.
◦ Published open evaluation benchmarks and state verification loops across complex multi-step reasoning tasks.
EDUCATION
• Ph.D. in Computer Science / Artificial Intelligence | Top Research University
• B.S. in Computer Science / Applied Mathematics | Leading Institute of Technology