Sasidharan B

AI Engineer with 1+ year of Internship experience passionate about building real-world AI products that automate workflows, improve decision-making, and solve practical business problems. Experienced in developing Agentic AI systems, multimodal document intelligence pipelines, and scalable RAG applications using modern AI frameworks and full-stack technologies. Able to quickly learn, prototype, and deploy production-oriented AI solutions in fast-paced environments.

Seeking opportunities to build scalable AI systems, contribute to innovative AI teams, and grow into an enterprise AI architect.

AI Engineer Intern – Generative AI at Pynevera Technologies Pvt. Ltd. (2025-12 – Present)

• Reduced manual receptionist handling effort by an estimated 90% by building a multilingual Voice AI-based Smart Receptionist system using Ultravox speech-native agents. Achieved ~98% successful appointment completion during internal prototype testing through Tamil and English multi-agent voice workflows.
• Built a multimodal PDF-to-knowledge-base pipeline using Google Gemini AI and ChromaDB for AI-powered patient query resolution. Automated complaint prioritization using Gemini AI and integrated into a real-time React dashboard. Improved clinic workflow efficiency during pilot deployment by integrating Supabase authentication, FastAPI backend services, and React-based staff operations dashboards.
• Currently working on deployment of the complete full-stack platform using Render and Vercel for scalable clinic operations management.

Project Trainee – Generative AI Use Case at Tata Communications Ltd. (2025-06 – 2025-08)

• Achieved ~99% structured extraction accuracy of 1000+ Enterprise PDF, Word Document by using Google Gemini 2.5 Flash instead of OCRs. Improved enterprise retrieval quality by implementing hybrid search architecture combining BM25 sparse retrieval and semantic vector search using ChromaDB. Increased response reliability through Agentic RAG + C RAG workflows that validated low-confidence responses and triggered intelligent re-retrieval.
• Reduced repeated LLM inference calls through semantic caching using LangChain-based semantic similarity matching. Designed modular multi-agent orchestration workflows using LangGraph for routing, retrieval, validation, and caching operations.

Software Developer Intern – R&D at Juzgodigital Pvt. Ltd. (2024-08 – 2025-05)

• Accelerated product feedback analysis workflows by an estimated 60% by building an automated audio sentiment analysis pipeline using OpenAI Whisper and GPT-based classification.
• Improved UI asset processing productivity by ~30% by developing an OpenCV-based RGB-to-HEX extraction tool for design workflows. Reduced production team dependency by creating technical documentation for Apache POI, Flask APIs, and SDK integrations.
• Reduced projected IVR infrastructure costs by ~25% through comparative benchmarking and analysis of Twilio and Plivo platforms. Maintained version-controlled engineering documentation across evolving APIs and SDK integrations.

B.Tech in Information Technology – United Institute of Technology (2022-11 – 2026-05)