Portrait of Aditya Raj Pandey

AI/ML @ CheQ · IIT Delhi '25 · Conversational AI for Fintech

Aditya Raj Pandey

Software Engineer·AI / ML

I build production-grade LLM and applied-NLP systems — from in-house entity recognition to retrieval-augmented assistants — and ship them end-to-end into fintech pipelines.

Bengaluru, India Open to opportunities
01 — About

Engineer at the intersection of LLMs, NLP, and product.

I'm a Software Engineer (AI/ML) at CheQ, building conversational AI and applied-NLP systems for fintech. After a B.Tech in Engineering Physics at IIT Delhi, I've focused on taking models from prototype to production — owning evaluation, compliance, and deployment, not just notebooks.

My core work spans information extraction, text classification, and LLM-powered systems in regulated, production environments. I like learning by building, asking a lot of questions, and the kind of thoughtful conversation that's usually about tech, philosophy, or a really good film.

  • EnglishFull Professional
  • HindiNative / Bilingual
  • FrenchElementary
0 F1 — production FinBERT NER
0 Merchant coverage (from 93%)
0 Sensor entries analysed
0 Movies scraped via ETL
02 — Experience

From prototype to production.

— Full-time experiencesince Nov 2025 · CheQ
— Experience incl. internshipscumulative across roles
  1. Software Engineer — AI/ML

    CheQ · Bengaluru

    Full-time Nov 2025 — Present

    Progressed from AI/ML Intern (Sep–Nov 2025). Building CheQWisor, an LLM-powered assistant for credit insights and user intelligence.

    Merchant Entity Recognition (in-house NER)

    • Led end-to-end development of an in-house Merchant NER, removing third-party vendor dependency in a regulated domain.
    • Benchmarked SpaCy, GLiNER and FinBERT; productionized FinBERT with evaluation and compliance validation.
    • Migrated label schema BIO → BILO and scaled data, lifting coverage from 93% → 99% at 97% F1.

    LLM Sentiment & Issue Classification

    • Scoped the problem with Product — defining taxonomy and ownership through cross-functional collaboration.
    • Designed prompt-based sentiment & issue classification with iterative prompt-engineering refinement.
    • Shipped user-sentiment intelligence via alerts and logging, enabling data-driven product decisions.

    Chatbot Web Retrieval (domain-specific crawling)

    • Ran a POC evaluating Firecrawl, Exa, Vertex AI and Crawl4AI for chatbot web-search integration.
    • Analysed 3 months of chatbot queries and curated a 50-query golden dataset across all query types.
    • Built a weekly crawl → LLM-summarization → vector-DB embedding pipeline for retrieval.
    • Python
    • Flask
    • FinBERT
    • SpaCy
    • GLiNER
    • LangChain
    • RAG
    • FAISS
    • Prompt Engineering
  2. AI Data Trainer

    Remotasks · New Delhi

    Internship Dec 2023 — Jan 2024
    • Rated AI model outputs and crafted ideal responses using coding skills to improve accuracy and relevance.
    • Developed and rated coding prompts, providing optimal solutions to improve model performance (~30 hrs/week).
  3. Operations Manager

    Bharat Skool · New Delhi

    Internship May — Jul 2023
    • Worked with the founder on YouTube content strategy to boost educational outreach.
    • Managed a team of IIT-JEE mentors and led marketing & sales efforts to grow the user base.
  4. Subject Matter Expert

    Snipe · New Delhi

    Internship Dec 2022 — Apr 2023
    • Curated educational content with a 5-person quality-control team.
    • Produced 40+ hours of structured, comprehensive lecture videos within one month.
  5. IIT-JEE Mentor

    Exphub · New Delhi

    Internship Jul — Aug 2022
    • Mentored aspirants and supported structured exam preparation.
03 — Projects

Things I've designed, built and shipped.

Email Automation Service

  • Automated personalized outreach at scale with template rotation and delays
  • CSV-driven contacts with duplicate-send prevention and auto PDF attachment
  • Web dashboard with logging; deployed on PythonAnywhere
  • Flask
  • Python
  • SMTP
  • CSV
  • PythonAnywhere

LLM-Powered SQL Chatbot

  • Converts plain-English questions into executable PostgreSQL instantly
  • Layered Redis cache → ChromaDB schema retrieval → LLM generation
  • Semantic caching cut repeat-query latency; React dashboard
  • FastAPI
  • PostgreSQL
  • Redis
  • ChromaDB
  • Groq / OpenAI / Claude
  • React

Cyclone Preheater Anomaly Detection

  • Unsupervised anomaly detection across 370k+ sensor readings, 3 years
  • Tuned DBSCAN via k-distance & silhouette scores
  • Surfaced pressure/temperature anomalies and localized process deviations
  • Python
  • DBSCAN
  • scikit-learn
  • Pandas
  • Time-series

Movie Box-Office Data Scraper

  • ETL pipeline scraping 3000+ movies from the-numbers.com
  • Parallelized with ThreadPoolExecutor for faster throughput
  • Cleaned, transformed, enriched with TMDB metadata into CSV
  • Python
  • BeautifulSoup
  • ThreadPoolExecutor
  • ETL
  • TMDB API
04 — Skills

A full-stack AI toolkit.

Languages & Core

  • Python
  • Java
  • SQL
  • Data Structures & Algorithms
  • OOP / LLD
  • Git & GitHub

Machine Learning & GenAI

  • PyTorch
  • TensorFlow
  • scikit-learn
  • Hugging Face Transformers
  • FinBERT
  • SpaCy
  • GLiNER
  • LangChain
  • LangGraph
  • RAG
  • LLM Orchestration (OpenAI / Azure / Claude)
  • Prompt Engineering
  • FAISS
  • ChromaDB
  • Sentence Transformers
  • NLP
  • Multi-Agent Systems

Data & Backend

  • FastAPI
  • Flask
  • PostgreSQL
  • Redis
  • NumPy
  • Pandas
  • Matplotlib
  • Seaborn
  • BeautifulSoup
  • ETL

Tools & Platforms

  • Streamlit
  • Power BI
  • Advanced Excel
  • Unstructured.io
  • Vercel
  • Neon
  • Upstash

Ways of Working

  • Cross-functional Collaboration
  • Stakeholder Management
  • Agile
  • System Design Thinking
  • Risk Assessment
  • Model Governance
05 — Education

Foundations.

B.Tech, Engineering Physics

Nov 2021 — May 2025

Indian Institute of Technology, Delhi

Engineering Physics with a steady pull toward computation, data and machine learning — the bridge into building AI systems professionally.

Certification

Transformer Models & BERT

Google Cloud

  • Transformers
  • BERT
  • NLP
06 — Achievements & Leadership

Beyond the codebase.

JEE Advanced 2021 All-India Rank 4326
JEE Main 2021 All-India Rank 6451
9 Leadership Roles Across IIT Delhi student bodies

Positions of Responsibility

Arts & Culture

Event Coordinator — Physics Society (CAIC)

IIT Delhi · Aug 2023 — Apr 2024

  • Ran Intern Insights for freshers, fostering community and knowledge sharing.
  • Led the Simulation Challenge and organised Professor–Student Meetups.
Arts & Culture

Logistics Coordinator — Rendezvous IITD

IIT Delhi · Aug 2023 — Apr 2024

  • Streamlined logistics for Comedy Hunt, Qawwali Night & SPIC MACAY.
  • Commanded a 10-person team across stationery to tech-equipment needs.
Science & Tech

Event Executive — TRYST

IIT Delhi · Jan — Apr 2023

  • Secured sponsorships from Quickheal & Neovel through pitching.
  • Ran 6 events with CXO-level collaboration; 700+ registrations.
Arts & Culture

Infrastructure Executive — Rendezvous IITD

IIT Delhi · Jan — Aug 2023

  • Coordinated stalls for Techno, Beardo, Havells & Swiss Beauty.
  • Curated live performances with 100+ artists; streamlined pronite food stalls.
Education

Marketing Executive — Hindi Samiti

IIT Delhi · Feb — Mar 2023

  • Marketed the MUN debate, reaching 500+ debaters across 20+ colleges.
Arts & Culture

Publicity Executive — Physics Society (CAIC)

IIT Delhi · Aug 2022 — Aug 2023

  • Drove publicity and engagement across the society's events.
Social Action

Hospitality Executive — Indradhanu

IIT Delhi · Dec 2022 — Feb 2023

  • Ran hospitality for 6 events over 3 days at VIBHINN, a queer-collective conference.
Dramatics

Writer / Actor / Assistant — Ankahi

IIT Delhi · Mar 2022

  • 2nd place in an inter-hostel short-filmmaking competition; 15-person team, 10-day build.
Dramatics

Actor — Ankahi

IIT Delhi · Jul — Aug 2022

  • Co-wrote, acted in and produced a 1971-war stage play for an audience of 1000, in 7 days.
07 — Contact

Let's build something.

Open to roles in AI / ML, GenAI and software engineering. The fastest way to reach me is email.