About Utkarsh
Production AI / ML Engineer - Generative AI, AI Agents, Prompt Engineering, LLMs (GPT / LLaMA / Gemini), RAG, Vector DBs (Pinecone), MLOps/LLM-Ops, Real-time STT/TTS, Telephony (Twilio, Plivo), LiveKit, Python, FastAPI, Flask, PostgreSQL (query/index tuning), Microservices, Docker, AWS/Azure/GCP, CI/CD. Built multi-tenant voice platforms (2,000+ concurrent calls across 13+ orgs), implemented RAG pipelines with 10.4× cold-start reduction, and achieved 90–95% query time improvements in production.
English
Native or bilingual
Remote only
Primarily works remotely
Experience
- Bigship TechnologiesAI EngineerJune 2025 - Today (1 year and 2 months)• • Drove end-to-end development of production-scale AI voice infrastructure using Python, orchestrating 500+ concurrent intelligent calls and 5,000+ daily calls, whilst architecting advanced LLM-powered conversational AI flows with real-time STT/TTS integration—achieving 80% response accuracy improvement.• • Engineered and optimized sophisticated multi-provider telephony pipelines integrating GPT-4, Deepgram STT, and ElevenLabs TTS through custom Python microservices, resulting in 45% latency reduction and seamless agent handoff capabilities, whilst implementing intelligent voicemail detection, DTMF support, and dynamic call routing to enhance enterprise-grade voice automation scalability.
- Scicom.aiAI Engineer (Freelance)October 2025 - December 2025 (2 months)• • Architected and deployed a production-grade realtime AI voice avatar system with Retrieval-Augmented Generation (RAG), integrating LiveKit Agents, Groq LLaMA-3.3-70B, Deepgram Aura TTS, and Beyond Presence avatars to enable real-time, enterprise-scale conversational AI for customer service workflows.• • Engineered and optimized an asynchronous RAG pipeline using Pinecone serverless vector database and Sentence Transformers embeddings, achieving 10.4x cold-start latency reduction (3.9s to 378ms), thread-safe concurrent session handling, and scalable knowledge ingestion for high-availability production deployments.
- ZuduBackend AI Voice Developer (Freelance)October 2025 - November 2025 (1 month)• • Architected enterprise-grade multi-tenant backend infrastructure integrating LiveKit with multi-provider telephony systems (Twilio, Plivo) to orchestrate 2,000+ concurrent AI voice calls across 13+ organizations, achieving 66% performance improvement and 40ms webhook latency.• • Engineered comprehensive analytics system with 11 RESTful API endpoints and automated credit management, whilst optimizing PostgreSQL database through 43 strategic indexes—resulting in 90-95% query time reduction and implementing JWT-based security, Chargebee integration, and Azure Blob archival.
Recommendations
Be the first to recommend Utkarsh
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Oracle Cloud Infrastructure 2025 Certified AI Foundations Associate2025Oracle Cloud Infrastructure 2025 Certified AI Foundations Associate
- "I built a virtual agentIBM2025"I built a virtual agent