Senior Manager – Agentic AI & ML Engineering - Mantra Softech India - India
(2026-04)
- Architected a four-agent BDR automation platform (Prospect, Similar Discovery, Intent, Research) on LangGraph orchestrating autonomous prospect identification end-to-end.
- Designed web-scraping RAG pipeline using LlamaIndex, LangChain, and ChromaDB to solve LLM context-window limits on bulk scraped web data.
- Built two-stage retrieval: BM25 lexical filter to discard junk chunks + BGE Rerank for semantic reranking of relevant context.
- Delivered 20+ daily qualified hot leads globally with Sales POV, Tech POV, and Management POV insights, flagging quality pinch points to the sales team.
- Built agentic automation system to process and reconcile uploaded invoices and POs within ERP workflow, eliminating manual data entry.
- Developed chatbot-driven analytics agent delivering on-demand data visualizations from natural-language employee queries.
- Defined and owned end-to-end agentic AI product architecture and engineering roadmap
- Architected LangGraph-based multi-agent orchestration framework from ground up
- Led product strategy for B2B sales automation and ERP intelligence platforms
- Guided junior AI engineers on RAG pipeline design, LangChain workflows, and agent debugging
- Collaborated directly with sales team translating AI lead-gen insights into outreach strategy
- Represented AI product capabilities in client-facing discussions and stakeholder demos
- Drove technical decision-making on stack selection (LlamaIndex, ChromaDB, BM25, BGE Rerank)
- Established engineering best practices for agentic system reliability and context management
Senior Manager - ML Infrastructure & Generative AI - Firstsource Solutions Limited - Mumbai, India
(2024-08 - 2026-04)
- Pioneered Multi-Agent Orchestration framework using CrewAI and AutoGen enabling complex decision-making across 5+ specialized AI agents.
- Architected end-to-end Agentic AI workflow for claims processing; reduced manual validation time by 85%.
- Designed self-healing agent systems with feedback loops achieving 98% autonomous completion rate.
- Implemented agent-to-agent communication protocols for seamless handoffs across extraction, validation, decision-making, and reporting agents.
- Architected enterprise-scale RAG-based IDP platform processing 5,000+ documents daily at 95% extraction accuracy, 80% fewer hallucinations.
- Designed 16-node NVIDIA V100 GPU cluster with PyTorch DDP achieving 4x throughput; 70% cost reduction through quantization.
- Achieved <2s API response times for 10K+ daily inference calls.
- Led advisory with 15+ C-level executives; delivered PoV-to-Production for 3 deployments ($2.5M revenue), accelerating time-to-value by 60%.
- Conducted business transformation workshops achieving 90% client satisfaction scores.
- Architected enterprise-grade multi-agent and RAG platforms end-to-end
- Owned ML infrastructure strategy across multi-cloud environments (AWS/Azure/GCP)
- Led C-suite advisory engagements across banking and insurance verticals
- Drove pre-sales technical presentations contributing directly to contract wins
- Managed cross-functional AI engineering teams and set delivery standards
- Oversaw GPU cluster design, LLM training pipelines, and deployment optimization
- Mentored senior and mid-level engineers on agentic frameworks and RAG design
- Defined organization-wide GenAI engineering best practices and tooling choices
- Governed PoV-to-production transitions, overseeing stakeholder alignment and delivery
Senior Technical Lead & Scrum Master - Accenture Advanced Technology Center - Gurgaon, India
(2021-10 - 2024-07)
- Deployed multimodal GenAI system using LayoutLM for banking deal analysis with 92% document understanding accuracy.
- Implemented custom BERT-based architectures with specialized tokenization for financial document processing.
- Delivered full MLOps pipelines with model versioning, A/B testing, and automated deployment serving 50K+ daily requests at 99.9% uptime.
- Architected containerized deployment using Docker and Kubernetes for scalable model serving.
- Conducted PoV presentations for 8 enterprise clients leading to 80% reduction in support workloads via LLM-powered conversational AI.
- Led knowledge transfer sessions training 30+ team members on LLM fine-tuning, RAG, and agentic workflow design.
- Led Agile development sprints as Scrum Master for AI/ML delivery teams
- Designed and implemented multimodal GenAI and MLOps solutions
- Mentored 30+ engineers on LLM fine-tuning, RAG, and agentic workflows
- Conducted technical PoV demos and solution scoping for enterprise clients
- Contributed to pre-sales technical workshops and client solution design
- Established MLOps standards: model versioning, A/B testing, CI/CD pipelines
- Collaborated across business and engineering teams to align AI delivery with client goals
Technical Leader - Computer Vision & ML - HCL Technologies - Noida, India
(2019-03 - 2021-10)
- Led architecture for real-time computer vision systems across 50+ edge locations with 7ms inference latency through advanced quantization.
- Designed distributed inference pipelines leveraging GPU acceleration with intelligent fallback mechanisms.
- Conducted solution design workshops with retail and manufacturing clients, architecting custom CV/ML solutions.
- Provided pre-sales technical expertise contributing to 5 contract wins worth $1.8M.
- Led CV system architecture across 50+ distributed edge sites
- Developed and optimized real-time inference pipelines with GPU acceleration
- Mentored junior engineers on model quantization and edge optimization
- Participated in pre-sales workshops and client solution design sessions
- Designed distributed deployment strategies with fallback mechanisms
- Collaborated with retail and manufacturing clients on problem scoping
- Reviewed code and guided team members on best practices for CV/ML systems
Image Processing & Machine Learning Engineer - Kisan Network - Gurgaon, India
(2018-07 - 2019-03)
- Developed Faster R-CNN-based object detection models for crop classification using TensorFlow.
- Designed analytics pipeline for object size measurement using reference-based detection algorithms.
- Independently developed and trained Faster R-CNN object detection models
- Implemented end-to-end TensorFlow ML pipelines for agricultural data
- Performed dataset curation, annotation, and augmentation for CV tasks
- Designed image processing workflows for field-level crop analysis
- Documented model training experiments and performance results
- Collaborated with domain experts to understand agricultural requirements
Senior Research Fellow - Defence Laboratory (DRDO) - Jodhpur, India
(2014-05 - 2018-05)
- Led research on Camouflage Object Efficacy Measurement Systems using advanced computer vision techniques.
- Developed prototype detection systems and conducted performance benchmarking under defense-grade constraints.
- Independently designed and implemented CV algorithms in C++ and Python
- Conducted experiments, analyzed results, and documented research findings
- Developed prototype systems for camouflage detection and measurement
- Performed literature surveys and contributed to research publications
- Collaborated with defense science team on algorithm development objectives
- Maintained lab equipment and managed research data pipelines
Android Developer - Tri-Force Consulting Services Pvt. Ltd. - Ahmedabad, India
(2013-04 - 2014-04)
- Developed enterprise Android applications using Java API 21 for client-facing business use cases.
- Integrated backend APIs and implemented UI/UX components for production-deployed apps.
- Developed and maintained enterprise Android applications from scratch
- Implemented UI/UX components and REST API integrations
- Debugged, tested, and resolved issues in existing mobile codebase
- Participated in daily standups, sprint planning, and code reviews
- Documented app modules and wrote unit test cases
Research Service Provider - DevBhoomi Institute of Technology - Dehradun, India
(2011-09 - 2013-03)
- Developed clustering algorithms for Mobile Ad-Hoc Networks (MANETs).
- Conducted network simulation experiments and documented algorithmic performance outcomes.
- Implemented clustering algorithms for MANET environments
- Ran network simulation experiments and analyzed results
- Conducted literature review and assisted faculty in research documentation
- Maintained research data and experimental records
- Collaborated with academic supervisors on research objectives
