Yash K- Senior Data Engineer

Major Projects

Automated Job Matching and Voice-Based Applicant Engagement Platform

Built a backend automation engine to streamline job scraping, nurse-to-job matching, and candidate communication using AI voice calls. Utilized FastAPI for scalable service orchestration, APScheduler for job scheduling and retries, and PostgreSQL for job-state persistence. Integrated Bland AI for real-time applicant voice interactions, Microsoft Graph API for scheduling interviews, and MixRank for lead enrichment. Automated email alerts and used background threading and webhooks to minimize recruiter workload and improve efficiency.

UK

Backend & Services:

APScheduler
FastAPI
PostgreSQL
Threading

AI & Integrations:

Bland AI for voice automation
Microsoft Graph API
MixRank API

DevOps & Tools:

Background threading
Email pipelines
Webhooks

Conversational AI System for Legal Document Understanding and Clause Extraction

Developed a legal document assistant that processes PDFs and images to extract summaries, entities, and contractual clauses through natural language interactions. Used FastAPI-based modular microservices and integrated OpenAI APIs, spaCy, and OCR for text extraction, NER, and prompt-driven clause detection. Added masking of sensitive information and dynamic checklist-based behavior to tailor model output. Tools like PyMuPDF, structured logging, PDF compression, and APScheduler helped ensure performance, security, and operational reliability.

UK

Backend & Services:

FastAPI
Modular Microservices

AI & NLP:

Entity masking
OCR
OpenAI API
Prompt Engineering
Spacy

DevOps & Tools:

APScheduler
PDF compression
PyMuPDF
Structured logging

3D Foot Modeling System from Smartphone Videos Using Computer Vision

3D Foot Modeling System from Smartphone Videos Using Computer Vision Engineered an AI-driven video processing pipeline to generate 3D foot models from user-recorded videos. Used YOLO for foot detection, TensorFlow for segmentation validation, and FastAPI to coordinate backend logic. Integrated RealityCapture API to convert extracted frames into accurate 3D plantar models. Enabled real-time user feedback during video recording and implemented cron jobs and automated retry logic for robust task reprocessing. OpenCV powered the measurement calculations and visual validation.

USA

Backend & Services:

FastAPI
Frame extractor
RealityCapture API

AI & CV:

TensorFlow
YOLO for foot detection

DevOps & Tools:

Automated retry logic
Cron Jobs
OpenCV

AI-Powered Reverse Image Search System for Facial Recognition

Created a facial search platform that compares an input image against a face database using vector embeddings and high-speed similarity ranking. Used InsightFace to generate face embeddings and FAISS for efficient similarity search based on cosine distance. Designed a FastAPI endpoint to handle image uploads and return matched results with rank scores. Deployed MongoDB for embedding storage and query handling, while optimizing backend logic for speed and accuracy in large-scale retrieval.

India