Hi, I'mAmit Kumar

Clients: Pfizer, Sanofi, Papa John's, GMC

• Built a scalable framework for custom chatbots with Flask, Scikit-learn, NLTK, SpaCy, NumPy and Hugging Face transformers.
• Developed a query pre-processing framework covering spell correction, semantic analysis, lemmatization, N-grams and word embeddings.
• Engineered a containment strategy by integrating the OpenAI Chat Completion API into fallback intents for out-of-scope queries.
• Built an XGBoost + Random Forest ensemble forecasting multi-region sales with 87% accuracy, cutting inventory holding costs by 12%.
• Designed an automated feature engineering pipeline (imputation, SMOTE, scaling, encoding) that reduced manual data prep by 70%.

Client: Novartis

• Architected a GenAI medical data retrieval platform with advanced RAG pipelines, improving data-driven decision-making efficiency by 60%.
• Integrated FAISS & CosmosDB vector databases as a scalable patient knowledge base, improving retrieval efficiency by 40%.
• Orchestrated LLM workflows with LangChain, LangSmith and LlamaIndex for accurate, context-aware responses.
• Fine-tuned the MedPalm2 LLM on patient-specific data to improve precision and grounding.

Client: Google

• Module Lead Developer for a GenAI deployment serving 18M+ CX sessions, sustaining 75% GenAI coverage with 3-5s latency.
• Optimised payment & billing workflows, lifting mobile payment success from 70% to 92% with an average of 1.39 attempts.
• Architected a multi-agent ecosystem on Google CX Agent Studio with 45 sub-agents and 61 tools.
• Led agent evals and hill-climbing with CXAS-SCRAPI and Google Antigravity, reducing error volumes by 38%.
• Google Advisory Expert for three external clients, building ADK + MCP + A2A multi-agent systems that automated 70% of manual work.