AI / ML Engineer · NTU Singapore

Aditya Kumar Goel

Responsible AI · LLMs · Computer Vision

I build trustworthy, production-ready GenAI systems — from data pipelines and RAG to safety evaluation. MSAI candidate at NTU, formerly Responsible AI Senior Analyst at Accenture.

Get in touch View work
30+
AI use cases evaluated
5+
GenAI solutions shipped
IEEE
Big Data 2024 author
Aditya Kumar Goel
Currently
AI Engineer @ Go-Genie
01 — Impact

Snapshot of focus & outcomes

A quick read on where I work and the results I drive.

Focus
Responsible AI, LLMs, CV

End-to-end GenAI solutions with built-in safety evaluation.

Industry
Pharma & Finance

Modernized risk reviews and built quantitative metrics.

Research
IEEE Big Data 2024

Numerical validation for financial summaries.

Delivery
Multiple GenAI PoCs

LLM apps taken from data to deployment.

02 — About

Building reliable AI

I'm an AI/ML engineer focused on trustworthy systems and production-ready delivery — translating research into workflows teams can actually adopt.

My work centers on LLM fine-tuning, retrieval-augmented generation, and evaluation pipelines, with a deep emphasis on Responsible AI and safety for high-stakes domains like pharma and finance.

I'm seeking collaborations and roles where reliability, governance, and measurable impact are core product requirements — not afterthoughts.

Who I am

An engineer who turns research into dependable, deployable systems.

What I specialize in

LLM fine-tuning, RAG, and evaluation pipelines — safety-first.

What I'm seeking

Roles where governance and measurable impact are first-class.

Artificial Intelligence

Responsible AI GenAI LLMs Computer Vision Vector Database Semantic Search Embeddings Machine Learning Deep Learning NLP RAG

Languages & Frameworks

Python PyTorch TensorFlow OpenCV FastAPI Hugging Face LangChain Scikit-learn Postman

MLOps & Cloud

CI/CD for ML Azure ML AWS SageMaker DVC Weights & Biases Model Monitoring Docker MLflow
03 — Experience

Where I've worked

Professional roles, research, and leadership.

Jan 2026 — Present
Go-Genie
Singapore

AI Engineer Intern

→Develop ML models with Scikit-learn and PyTorch for logistics demand forecasting, route optimization, and load balancing.
→Engineer data pipelines and integrate AI inference into backend microservices via REST APIs for scalable deployment.
→Design a dynamic RAG pipeline using Gemini, LangChain, and BM25, with sequential API triggering and context-aware prompting.
→Built an AI-powered HRMS chatbot for leave management, payslips, and personalized FAQs — cutting administrative overhead.
Jul 2022 — Jul 2025
Accenture
Bengaluru, India

Responsible AI Senior Analyst

→Evaluated 30+ AI business use cases and collaborated with EU representatives to interpret the AI Act, creating self-service evaluation workflows and compliance reports.
→Delivered 5+ end-to-end GenAI solutions, including automating knowledge management with RAG to cut client time spent on database lookups.
→Developed a Computer Vision pipeline using Tesseract OCR to extract and structure invoice data for downstream RAG.
→Built a quantitative Responsible AI evaluation system handling thousands of batch requests with auto-prioritization.
→Applied MLOps best practices across Azure (CI/CD) and AWS (EC2, ECS, SageMaker, Bedrock, S3) for training, versioning, and deployment.
Mar 2024 — Mar 2025
OP
OP Venture Cooperative
Remote

CO Venture Fellow · Cohort 6

→Participated in venture capital education and investment processes, supporting due diligence and early-stage founders.
Oct 2020 — May 2021
VO
VentureOps
India

President

→Hackathon Lead for 'Hello World 4.0' (Oct 2020).
→Built the club website: venturesityvit.github.io.
Aug 2019 — Feb 2020
Centre of Excellence for Autonomous Vehicles Research
India

Student R&D Engineer

→Developed computer vision for an autonomous EV.
→Built optical flow for vehicle direction estimation.
04 — Research

Published research

Conference Paper IEEE Big Data 2024

Factual Accuracy Checking by Validating Numerical Values in Financial Summaries

Aditya Kumar Goel et al. · 2024

Validates numerical accuracy in financial summaries through automated checks and transformer-based reasoning — reducing factual errors in generated financial reporting.

View paper Google Scholar
05 — Projects

Featured projects

Innovation across the AI/ML stack — from product to research.

Graphics · WebGPU Featured

Walking inside a DeepSDF latent space

A real-time WebGL / WebGPU visualization that flies through a learned DeepSDF latent space — morphing smoothly between 3D shapes to make a neural shape representation tangible and explorable.

WebGPUWebGLDeepSDF3DLatent Space
Read on LinkedIn
Product 01

Doorlah

A platform focused on better discovery and engagement for local services — designed for fast navigation and clear calls to action.

OutcomePublic launch with a live site experience.

WebProductUX
Visit
NLP 02

Product Market Profiling Dashboard

A product-focused sentiment engine using NLP key-phrase extraction and source filtering across social media and news, with a high-performance FastAPI backend on Groq LPU and async orchestration.

OutcomeNear real-time insight delivery for stakeholder dashboards.

PythonFastAPIGroq LPUOpenAI
Finance + NLP 03

Stock Prediction

A pipeline combining market news signals and price features to forecast short-term equity direction, evaluated with rolling-window validation and daily prediction snapshots for decision support.

OutcomeFaster sentiment snapshots for market context.

PythonNLPFinance
GitHub
Responsible AI 04

Bias & Toxicity Detection

Fine-tuned transformer models with Hugging Face for multi-label toxicity and bias classification in AI-generated text, plus a synthetic data pipeline to study and mitigate adversarial attacks.

OutcomeImproved safety review coverage for GenAI outputs.

PyTorchHugging FaceTransformers
Document Intelligence 05

GenAI Q&A with Multi-Doc Extraction

A RAG pipeline integrating OCR extraction, vector indexing, and semantic search for multi-document retrieval, with dynamic prompting and GPT-3.5 Turbo summarization — cutting token cost by 80%.

OutcomeAccelerated multi-document lookups for analysts.

PythonLangChainOCRVector DB
Computer Vision 06

Manufacturing Defect Detection

CNN-based vision models trained in PyTorch for vehicle manufacturing defect detection, deployed via an OpenCV pipeline to automate quality checks on production lines.

OutcomeHigher QA throughput through automated screening.

PyTorchOpenCVDeep Learning
06 — Education

Academic foundation

M.Sc. in Artificial Intelligence
Nanyang Technological University, Singapore
Aug 2025 — Jun 2026

CGPA 3.9 / 5. Advanced coursework in LLMs, trustworthy AI, and applied research.

B.Tech. in Information Technology
Vellore Institute of Technology
Jul 2018 — Jun 2022

CGPA 4.1 / 5. Special Achiever's Award 2020–21.

07 — Awards

Awards & honors

Accenture
ACE Inspiring Innovator
Accenture
Innovation Excellence
Accenture
Technical Excellence
Accenture
Academia
HPAIR Delegate
Harvard College Project for Asian & International Relations
Special Achiever's Award
VIT University
External
Top 20 · Prospect 100
Global Tech Innovation
08 — Beyond work

Writing & photography

Thoughts on Medium, and moments captured on Instagram.

Medium @adityagoel1999

Loading latest post…

Fetching from Medium

Hang tight while the latest story loads.

Visit Medium
View on Instagram

09 — Contact

Let's build
something

Open to AI/ML research and product roles starting in 2026. Always happy to talk about trustworthy AI.

adityaku003@e.ntu.edu.sg
LocationNTU, Singapore
Phone+65 8846 5661
StatusAvailable 2026