Akshay Kumar C P

AI Engineer Lead (Manager)

Summary

• Architecture Leadership: Designed and led development of enterprise-scale AI platforms using microservices architecture with 10+ backend services and module federation frontends.
• Team Management: Led cross-functional teams of 5+ engineers (AI, Backend, Frontend, Data Engineering) through complete SDLC from design to production deployment.
• Technical Innovation: Architected production-grade RAG, Multimodal-RAG, and Agentic AI systems using Langchain, Langraph, Claude Sonnet 4.5, GPT-4o, and AWS Bedrock.
• Cloud & DevOps: Implemented CI/CD pipelines with GitHub Actions, deployed scalable solutions on Azure (Container Apps, Static Web Apps, Redis, Storage) and Databricks.
• Business Impact: Delivered 1.4M€ annual cost savings through AI automation, achieved 60% accuracy in compliance systems, improved ML model performance by 20%.
• Strategic Consulting: Led requirement gathering, RFP analysis, solution architecture design, and stakeholder alignment for enterprise AI initiatives.

Project's

Business4AI - Enterprise AI Platform

• Role: AI Engineer Lead - End-to-end ownership from architecture to production deployment
• Architecture: Designed scalable microservices platform with 10 backend services and 4 module federation frontends
• AI Integration: Implemented multi-LLM strategy using Claude Sonnet 4.5 (via Claude Code CLI) and AWS Bedrock for intelligent automation
• Cloud Infrastructure: Azure-native deployment - Static Web Apps for frontend, Container Apps for services, Redis for caching, Blob Storage for data persistence
• DevOps Excellence: Built automated CI/CD pipelines with GitHub Actions for continuous deployment and testing
• Tech Stack: Python (FastAPI), React.js, Azure Services, Claude Sonnet 4.5, AWS Bedrock, Docker, Redis
• Impact: Enabled organization-wide AI adoption with production-ready, scalable platform serving multiple business units

Multimodal-RAG Playground - AI Document Intelligence Platform

• Solution: Built enterprise platform processing multimodal content (Images, Video, Audio, PDFs with embedded images) for intelligent Q&A
• Leadership: Led cross-functional team of 8+ (UI, Backend, Data Engineering, AI) delivering features across sprints
• Architecture: Designed RAG pipeline with vector embeddings, semantic search, and context-aware retrieval using Databricks and Azure services
• DevOps: Configured Azure DevOps CI/CD for automated testing, deployment, and sprint releases
• Tech Stack: Langchain, Azure OpenAI, Databricks, Azure Cognitive Services, Vector DBs
• Impact: Enabled intelligent document analysis at scale with 80%+ accuracy in multimodal content extraction

Agentic AI Chatbot - Intelligent Reasoning System

• Innovation: Architected production-grade Agentic AI system with ReAct pattern implementation from scratch, enabling autonomous reasoning and tool use
• Features: Dual-mode interface (text + visualization), function calling for external integrations, context-aware multi-turn conversations
• Team Leadership: Led team of 5 engineers through design, development, and deployment phases
• Production Deployment: Deployed on Azure App Service with secure architecture using Key Vault for secrets, MySQL for state management
• Tech Stack: Langraph (agentic orchestration), FastAPI, GPT-4o, Azure App Service, Key Vault, Blob Storage, MySQL
• Impact: Reduced manual query resolution time by 70% with intelligent autonomous agent workflows

RAG Playground - Document Intelligence Engine

• Solution: Built intelligent document processing pipeline with custom parsers for 15+ file formats and AI-powered enrichment
• Architecture: Designed end-to-end RAG system with chunking strategies, vector embeddings, and semantic search optimization
• Innovation: Developed custom document enrichment modules using NLP for metadata extraction, entity recognition, and content summarization
• Tech Stack: Langchain, Azure OpenAI embeddings, Databricks Vector Search, Azure Cognitive Services
• Impact: Improved downstream chatbot accuracy by 35% through enriched vector database with enhanced context retrieval

RAISE Platform - AI Governance & Security

• Solution: Designed enterprise AI governance platform with human-in-the-loop monitoring, security controls, and quality assurance
• Security: Implemented prompt injection detection, input sanitization, and output validation to secure GenAI applications
• Quality Metrics: Built comprehensive evaluation framework for data quality, model performance, and response reliability
• Agent Architecture: Leveraged Microsoft Autogen for multi-agent orchestration and collaborative AI workflows
• Tech Stack: Microsoft Autogen, LLM APIs, Azure Security Services
• Impact: Established production-ready AI governance framework adopted across 10+ enterprise AI projects

Legal Tender AI Review - Compliance Automation

• Business Problem: Manual legal contract review taking 100+ hours per tender, high risk of compliance violations
• Solution: Built AI-powered contract validation system using RAG architecture with compliance rule engines and risk assessment
• Approach: Led end-to-end solution - requirement gathering, architecture design, prompt engineering experimentation, production deployment
• Tech Stack: GPT-3.5-Turbo, Azure OpenAI Embeddings (text-embedding-ada-002), Azure Cognitive Search (vector store), Langchain orchestration
• Business Impact: - 1.4M€ annual cost avoidance - 60% accuracy in high-confidence criticality assessment - 90% reduction in review time from 100+ hours to 10 hours

Code Assistant - AI-Powered Development Tool

• Solution: Built intelligent code generation system converting business requirements to production-ready PySpark code
• Constraint: Air-gapped environment requiring fully on-premise deployment without external API calls
• Innovation: Deployed open-source LLMs (Mistral, Llama 2) with quantization optimization on Databricks GPU compute for cost-effective inference
• Architecture: RAG-based system with code examples in vector DB, prompt engineering for code quality, Langchain orchestration
• Tech Stack: Databricks GPU, HuggingFace models, Quantization (4-bit), ChromaDB vector store, Langchain, Ollama runtime
• Impact: 50% reduction in data pipeline development time, enabled non-technical users to generate complex PySpark transformations

Generate Synthetic Data using Generative AI

• Tabular: Utilized libraries such as CTGAN, SDV, etc for generation.
• Images: Generative models such as GAN, BIG-GAN, stable diffusion, etc.
• Text: Used GPT-2, Mosiacml, Bloom, falcon, etc. for text generation tasks.
• Implemented using Hugging Face, TensorFlow Hub and other frameworks.

Personalized Marketing

• Personalize images using Open-Source Stable Diffusion Models.
• Performed Fine Tuning using Dreambooth Framework.

Natural Language to SQL

• Implemented using Open AI “text-davinci-003” model and Prompt engineering techniques.
• Built UI using Gradio framework for quick demo.

AWS Clean Room

• Take marketing decision without third party cookies.
• Configured collaboration between advertiser and publisher using, S3 and Glue.
• Amperity CDP for Identity resolution (IDR) and customer campaigns
• Technologies: Amperity Platform

Customer Satisfaction Analytics

• Research & analysis on requirements and planning.
• Brainstorming the solution with respect to requirements.
• Generating actionable insights from data and creating reports, presentation and dashboards to take key business decision.
• Perform analysis namely predicting loyal customers using cluster models, CSAT, NPS, SERVQUAL, Recency Frequency Monetary (RFM), Cohort analysis.
• Use python for automation (OOPS, REST APIs, logging, exceptional handling, config file. linting, etc.).
• Performed Sentiment classification and Topic Modelling as part of Satisfaction Metrics
• Technologies: Python & Machine Learning.

Hackathon

Insect Acoustic AI

• Is a Multi class audio classification. Extracted features from audio data and experiment with multiple model architectures such as ANN, Hugging Face pretrained models and fine tuning.
• Experimentation was performed on cloud AWS Sage maker.
• Evaluation metrics such as True Positive, Accuracy, etc.

Google Generative AI

• Theme: Customer products and retail: Personalized Copy Generate.
• Used Google's LLM "text-bison (latest)" model from Google Vertex AI .
• Personalized content generation for e-commerce platform such as:
• Create product description for new/existing products.
• Personalized content generation for promotional activities (SMS/Email/Social media) based on customer past transaction details.
• Come up with strategies for retaining churning customers.

Software Development Engineer - Machine Learning

Summary

• ML Engineering: Built production ML models for open-source license detection and copyright extraction, improving product accuracy by 20%.
• Automation & DevOps: Automated manual workflows using Python, REST APIs, MongoDB - reducing operational time by 60%.
• Product Engineering: Debugged and optimized internal products, delivered customer-critical fixes with 99.5% uptime SLA.
• Innovation: Researched and integrated AI/ML solutions including NLP-based chatbots and component analysis systems.

Project's

Open-Source License Detection & Copyright Extraction

• Business Impact: Increased component license mappings from 35% to 55% (+20% improvement) for enterprise software compliance
• Solution: Built NLP-based ML system for automated license detection and copyright holder extraction from source code
• Full-Stack ML: End-to-end ownership - data preprocessing, model training (Spacy NER), REST API development, production deployment
• Production Architecture: Deployed on UWSGI server with MongoDB for persistence, serving 10K+ daily predictions with 99.5% uptime
• Tech Stack: Python, Spacy (NLP), Scikit-learn, REST APIs, UWSGI, MongoDB

Q&A Chatbot for HR - Conversational AI

• Research & Evaluation: Comparative analysis of 4 chatbot platforms (Google Dialogflow, Amazon Lex, Azure LUIS, Rasa X) for production deployment
• Solution: Built intelligent HR chatbot with NLU, FAQ automation, and email integration for 500+ employees
• Architecture: Python backend with MongoDB for conversation history, multi-platform integration testing
• Business Impact: 70% reduction in HR ticket volume, 50% faster employee query resolution
• Tech Stack: Python, Dialogflow/Lex/LUIS/Rasa, MongoDB, Email APIs

Automation Tools, Python automation projects.

• Rules for identifying components.
• Download, extract important information from releases of component.
• Validate URL of component.
• Generate SQL queries, etc.

Master Of Computer Application (MCA)

CGPA: 7.91

Web Developer Intern

• Developed and maintained web applications using modern web technologies
• Implemented operational change management workflows and frameworks
• Optimized key performance indicators for web platform efficiency
• Gained hands-on experience in full-stack web development and agile methodologies