Rakesh Kumar - AI Product Manager

AI Product Manager & AI Lead

7+ years building agentic AI, RAG platforms, data products, and high-impact customer experiences across healthcare, e-commerce, and IT.

View Case Studies Download Resume LinkedIn

Product Impact

70%+

Manual tasks reduced using agentic AI workflows

80%

Faster insight generation with BI dashboards

10×

Clinical research turnaround acceleration

85%

ML + LLM diet engine accuracy

20k+

Users onboarded in 90 days (coach-driven wellness)

$50k+/mo

Incremental revenue via recommendations

3×

Experimentation velocity (AI playground)

1M INR

GMV leakage prevented per month

Professional Experience

  1. Feb 2023 – Present
    🏥

    AI Product Manager & AI Lead — Dr. Reddy's (Svaas Wellness)

    • Built agentic AI framework to automate workflows and fine-tune LLMs; reduced manual tasks by 70%.
    • Architected RAG pipeline for clinical intelligence with customized data cleaning and ingestion; 5% better retrieval score than Azure managed services.
    • Designed Deep Research Agent for clinical studies; cut research turnaround time by 10×.
    • Architected GenAI voice agent for clinical support with <4s latency enabling real-time interactions.
    • Developed internal AI Playground for rapid product and design teams; 3× faster prototyping and AI experimentation.
    • Engineered ML + LLM-based diet engine achieving 85% accuracy.
    • Integrated RAG agent into BI dashboards; 80% faster insight generation and self-serve analytics.
    • Implemented symptom analyzer and health risk assessment tool impacting 5k+ users in 4 months.
  2. Dec 2021 – Jan 2023
    🏥

    Product Manager (Customer Success, New Initiatives, Platform) — MediBuddy

    • Built tool generating valuable customer feedback; improved adoption and messaging.
    • Scaled B2B CS support and automated systems; raised CSAT from 2.8 to 4 and first contact resolution by 10%.
    • Launched social media automation; boosted organic reach.
    • Accelerated refunds with all payment sources; automated claims with third parties and insurers.
  3. Jul 2020 – Dec 2021
    👗

    Data Product Manager — eShakti

    • Improved conversion by 7% for new users via enhanced product discovery funnel.
    • Increased customization usage by 9% with $30,000+ per month CLV uplift.
    • Improved onsite search and sorting; 5% overall sales increase and $15,000/month reduction in overstock.
    • Built recommendation engine yielding $50,000+/month incremental revenue and 8% retention improvement.
  4. Jul 2015 – Apr 2018
    💻

    Software Engineer — Capgemini India Pvt. Ltd.

    • Resolved 10+ post-production issues; stabilized a web application.
    • 30% reduction in waiting time with 15+ new features enhancing UX.
    • Cut lead time of an upload service by 60% and reduced report generation time by 50%.
    • Developed and trained ML models using RNN, NN, and CNN for training programs.

Selected Case Studies

Agentic AI Platform — Healthcare

AI Agents LLM Automation

Problem

Manual processes and slow iteration for AI features across teams.

Action

Built a reusable agentic AI framework, custom data pipelines, and evaluation harness; integrated into BI stack.

Impact

70% reduction in manual tasks; 3× faster prototyping; 80% quicker insight generation.

Business Case

Addressed inefficiencies in healthcare workflows by leveraging AI to automate repetitive tasks, enabling faster decision-making and resource allocation for clinical teams.

PRD Highlights

  • Core Features: Agentic workflows, LLM fine-tuning, RAG integration for BI dashboards.
  • Success Metrics: Task reduction >70%, insight generation time <20% of original.
  • Tech Stack: Python, LangChain, custom RAG pipelines.
  • Stakeholders: Product, Design, Clinical teams.

Medical Deep Research Agent

RAG Knowledge Graph Multi-Agent SLMs

Problem

Pharma teams needed comprehensive secondary research on molecules covering epidemiology, competition landscape, clinical trials, standard of care, pharmacokinetics, safety profiles, and unmet needs—across fragmented public and private data sources.

Action

  • Built RAG pipeline ingesting both private enterprise data and public sources (PubMed, ClinicalTrials.gov, regulatory filings)
  • Constructed domain-specific Knowledge Graph for entity relationships (molecules, diseases, treatments)
  • Multi-agent architecture: specialized SLMs for entity extraction, TinyLLM for knowledge synthesis
  • Custom evaluation harness for retrieval accuracy and factual grounding

Impact

10× faster research turnaround with 95%+ retrieval accuracy, enabling faster time-to-decision in drug development.

PRD Highlights

  • Core Features: Multi-source RAG, Knowledge Graph traversal, trained SLMs, TinyLLM orchestration
  • Success Metrics: Turnaround time <10% of manual, retrieval accuracy >95%
  • Tech Stack: Vector DB, Neo4j Knowledge Graph, custom embeddings, LangGraph agents
  • Stakeholders: Medical Affairs, R&D, Regulatory

View Architecture →

Recommendation Engine at Scale — eCommerce

SVD Neural Networks Collaborative Filtering Dimensionality Reduction

Problem

Low product discovery rates and high inventory overstock in fashion e-commerce, leading to poor conversion and wasted stock.

Action

  • Built hybrid recommendation engine: User-User, Item-Item, and User-Item collaborative filtering
  • SVD-based matrix factorization with multi-layer architecture for embedding learning
  • High-dimensionality reduction techniques (PCA, t-SNE) for feature compression
  • Neural network layers for prediction refinement and personalization
  • A/B testing framework for continuous experimentation

Impact

+$50k/month incremental revenue, +7% new-user conversion, −$15k/month overstock reduction, +8% retention improvement.

PRD Highlights

  • Core Features: Hybrid SVD + neural net recs, real-time scoring, A/B testing
  • Success Metrics: Conversion uplift >5%, revenue increment >$50k/mo
  • Tech Stack: Python, TensorFlow, Spark, Redis for real-time caching
  • Stakeholders: Marketing, Merchandising, Data Science

Other Projects

GenAI Voice Agent

Architected low-latency voice agent for real-time clinical support interactions.

GenAI Voice Healthcare

Impact: Enabled seamless support with <4s latency.

ML + LLM Diet Engine

Engineered hybrid engine for personalized diet recommendations with 85% accuracy.

ML LLM Wellness

Impact: Improved user health outcomes in wellness platform.

Symptom Analyzer & Risk Assessment

Implemented AI tool for health risk evaluation, reaching 5k+ users in 4 months.

AI Health Tech

Impact: Enhanced early detection and user engagement.

AI Playground

Low-code platform enabling business owners to create prototype AI agents with just a few clicks. Features automated RAG pipeline creation, evaluation frameworks, and orchestration layer.

  • Database integration hooks for internal enterprise systems
  • Configuration layer for non-technical users to validate model responses
  • Addresses non-deterministic agent behavior with proper testing & validation

RAG LLM Evals Orchestration Low-Code

Impact: 3× faster prototyping and AI experimentation for product and design teams.

View Architecture →

SemaAI Agent

Fully on-premise enterprise chatbot built for data-sensitive environments with no external API dependencies.

  • TinyLLM for lightweight, fast inference on-prem
  • LoRA fine-tuning for domain adaptation without full model retraining
  • RAG-grounded responses for factual accuracy
  • Custom MLP layer for intent classification before LLM routing

TinyLLM LoRA RAG On-Prem Intent Classification

Impact: Secure, low-latency AI assistant for internal enterprise use cases.

Skills

Education

Certifications

IBM AI Product Manager, IBM AI Developer, Google Digital Marketing, Six Sigma Green Belt, LLM Courses.

💬 Ask Me Anything

Have questions about my experience, projects, or background? Chat with my AI assistant!

Contact

Email Me LinkedIn Back to top ↑