Full Stack AI Engineer Building production applications that combine machine learning, real-time processing, and scalable backend systems.

Currently working at 360labs.ai on applied AI features, system integration, and full stack product development in live environments.

Primary interests:

• Applied machine learning in production systems
• Retrieval and search architectures
• Speech and language processing
• Real-time application infrastructure

Model training • NLP • embeddings • inference pipelines

Service design • REST APIs • real-time communication • type-safe architecture

Interactive UI • data-driven interfaces • type-safe components

Persistence • deployment • environment management

Working on production AI and full stack systems deployed in real user environments.

Responsibilities include:

• Implementing backend services for AI-driven product features
• Integrating machine learning inference into application workflows
• Developing real-time communication and streaming functionality
• Designing APIs for model interaction and structured data processing
• Managing data storage and retrieval pipelines
• Supporting deployment, monitoring, and performance debugging
• Contributing to modular architecture and scalable feature development

https://vyu-gen.vercel.app/home

Semantic search system for structured academic question datasets.

System components:

• Text embedding generation
• Similarity-based ranking
• Topic filtering
• Dataset ingestion and indexing

Built as a full stack application with ML-backed retrieval.

Notes:
Demo runs in a limited compute environment and does not represent full system performance.

https://voclize-demo.vercel.app/

Streaming speech processing system that measures speaking behavior using quantifiable metrics.

Core functionality:

• Real-time speech-to-text
• Speaking rate estimation
• Pause detection
• Filler word tracking
• Session-level fluency scoring

Designed for structured communication analysis rather than subjective evaluation.

Notes:
Demo operates under restricted API usage and limited streaming duration.

Similarity-based recommendation using feature distance and clustering.

DistilBERT fine-tuned for text polarity prediction.

LinkedIn
https://www.linkedin.com/in/udayjit/

Email
Udayjit065@gmail.com