Headline

AI Engineer @Aptagrim | Pixels, PDFs & a bit of voice | Making weird things work

About

I’m an AI Engineer at Aptagrim, currently building BidPlan AI, an AI-powered bid and proposal management platform designed to help teams work through complex RFPs, long tender documents, requirements, risks, and proposal workflows. My work sits at the intersection of AI engineering, product development, and full-stack execution. I design AI workflows using LangGraph, build API-driven backend systems, work with PostgreSQL for structured data flows, and create React-based interfaces for product experiences. I also use AI coding tools like Codex heavily in my development process, combining fast iteration with proper system design. A large part of my current focus is turning messy, document-heavy business processes into structured AI workflows: intake, extraction, requirement mapping, risk analysis, proposal drafting, validation, and end-to-end flow orchestration. My earlier background is in deep learning and computer vision, including research work around YOLO-based medical imaging systems. That gave me a strong foundation in ML, model pipelines, and applied AI problem-solving. I’m interested in building practical AI systems that go beyond demos, products where agents, retrieval, APIs, databases, and user workflows come together to solve real operational problems.

Location

Hyderabad

AI stack

claude code

claude

hugging face

langchain

langgraph

openai

Work experience

AI Engineer at Aptagrim Limited, Feb 2026 - Present · 9 mos

AI Engineer at The EleFit Store, Jun 2025 - Jan 2026 · 8 mos, - Designed and shipped an AI meal-planning assistant using prompt engineering and FastAPI, adding SSE-based streaming so responses felt noticeably faster to users. - Refactored the LLM workflows behind customer queries, tightening structured JSON outputs and cutting down on malformed responses that needed manual fixing. - Partnered with product to turn recurring support questions into prompt-driven automations, reducing repetitive manual handling for the support team.

Freelance Data Scientist at Freelance, Nov 2024 - May 2025 · 7 mos, Worked with multiple companies on diverse technical problems, including: - Developed algorithms to extract jute fibre metrics from microscopic images using classical computer vision and deep learning techniques. - Built models to detect fetal head circumference in ultrasound scans, improving measurement accuracy. -Designed and implemented an interface to convert text-based conversations into real-time vocal interactions using speech synthesis and recognition.

Machine Learning Intern at Akridata, Jan 2024 - Sep 2024 · 9 mos, Bengaluru, Karnataka, India, - Worked on text prompt refinement for improved object detection while leveraging transformer-based models like OWLViT, enhancing Akridata’s Data Explorer with automated, high-quality labeling. - Built a chatbot to query pandas dataframes via natural language, improving command generation through chain-of-thought prompting. - Designed lightweight tracking solutions for video datasets, boosting object detection accuracy and user efficiency in Autolabel.

Deep Learning Research Intern at Indian National Centre for Ocean Information Services (INCOIS), Dec 2022 - Jan 2023 · 2 mos, Hyderabad, Telangana, India, Worked on a project aimed to find potential fishing zones in Bay of Bengal. Performed image segmentation using YOLO based model on satellite images. Got a testing mean average precision of 94% on test data. Used about 1700 images for training and testing

Education

Bachelor of Technology - BTech, Electronics and Instrumentation, Manipal institute of technology

Public links

github: https://github.com/sankalp1001