Skip to content
View FieryDeveloper's full-sized avatar
  • North Carolina

Highlights

  • Pro

Block or report FieryDeveloper

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
FieryDeveloper/README.md
Typing SVG

Β ChapelΒ Hill,Β NCΒ  Β guam@unc.eduΒ 




▍ whoami

name:         Amogh Gupta
role:         MS in Computer Science, UNC Chapel Hill  (Aug 2025 – May 2027)
research:
  - Prof. Snigdha Chaturvedi   β€”  LLM Alignment
  - Dr.   Neil  Gaikwad        β€”  Socially Responsible AI
teaching:
  - TA    Β· DATA 520           β€”  Research Methods for Socially Responsible AI
  - ASPSA Β· Athlete Student Tutor  β€”  Math, COMP, DATA 110
undergrad:    B.Tech CSE, IIIT Pune  Β·  CGPA 9.22  Β·  top 3 / 175
honors:       Honors in ML β€” 1st in batch (CGPA 9.75)  Β·  sole 10/10 SGPA, sem 7

▍ Featured Β· accepted at FAccT '26




Compounding Disadvantage: Auditing Intersectional Bias in LLM-Generated Explanations Across Indian and American STEM Education

A. Gupta, N. Patil, S. Ghosh, S. Gaikwad Β· arXiv:2601.14506

We audit how LLMs generate STEM explanations across language, caste, and national context β€” and find that disadvantages don't just add, they compound. Indian lower-caste learners asking in regional languages receive systematically lower-quality explanations than any single axis of bias would predict.

Read on arXiv Cite


▍ Research focus

                                    β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
                                    β”‚   RESPONSIBLE AI  /  NLP β”‚
                                    β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                                                 β”‚
          β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
          β”‚                          β”‚           β”‚           β”‚                          β”‚
          β–Ό                          β–Ό                       β–Ό                          β–Ό
  β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”        β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”         β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”         β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
  β”‚  MULTILINGUAL β”‚        β”‚   AUDITING    β”‚         β”‚   AI POLICY   β”‚         β”‚   RED-TEAM    β”‚
  β”‚   ALIGNMENT   β”‚        β”‚  INTERSECTNL  β”‚         β”‚   SYSTEMS     β”‚         β”‚  & SAFETY     β”‚
  β”‚               β”‚        β”‚     BIAS      β”‚         β”‚               β”‚         β”‚               β”‚
  β”‚ post-training β”‚        β”‚ lang Β· caste  β”‚         β”‚  4000+ docs   β”‚         β”‚  jailbreaks   β”‚
  β”‚ + inference   β”‚        β”‚  Β· country    β”‚         β”‚  knowledge    β”‚         β”‚  across 6+    β”‚
  β”‚ manipulation  β”‚        β”‚  FAccT '26 βœ“  β”‚         β”‚    graph      β”‚         β”‚   LLMs        β”‚
  β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜        β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜         β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜         β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
︎︎ Expand active projects

β‘  Multilingual Alignment of LLMs β€” Aligning model behavior and reasoning across languages to reduce cultural bias. Post-training + inference-time manipulation at different layers of the stack.

β‘‘ AI Policy & Engineering Education β€” A knowledge-graph-backed chatbot over 4,000+ policy documents spanning jurisdictions worldwide. Simulating and enforcing AI policy decisions. Studying how AI exposure shapes engineering students' learning patterns. Demo available on request.

β‘’ Language, Caste & Context in AI Explanations β€” Demographic disparities in AI-generated explanations across Indian & American STEM education. β†’ accepted at FAccT 2026.


▍ Experience Β· the through-line is applied AI

Jan – JunΒ 2025
Mumbai
Nomura Β· Software Engineering Intern
Shipped an internal budgeting & financial-planning platform used across teams. Closed 40+ JIRA tickets in 6 months β€” perf, bugs, UX. React front-end, Spring Boot services, Jenkins CI/CD, SQL, Agile.
Aug – DecΒ 2024
California Β· Remote
AIAssistant (Startup) Β· Backend Intern
Voice & agentic AI. Built client-demo prototypes on FastAPI + Twilio + Deepgram + ElevenLabs β€” call summarization, healthcare assistants, lead-gen voice bots.
May – AugΒ 2024
Pune
Ernst & Young Β· Data & GenAI Intern
GenAI solutions for Fortune-100 clients. Built a RAG pipeline over 10,000 docs and engineered metadata filters that narrowed retrieval from 10,000 β†’ 50 candidates β€” a 200% search-efficiency lift.
Apr – AugΒ 2024
IIT Kharagpur
IIT Kharagpur Β· Research Intern β€” Prompting & LLMs
Red-teamed ChatGPT, Gemini, Nous-Hermes, Mixtral-Instruct, Cohere Coral & more using custom multilingual prompts. Cross-evaluated jailbreak outputs with multiple HuggingFace sentiment models. Wrote automated Python eval harnesses.
Jan – AprΒ 2024
IIT Indore
IIT Indore Β· Research Intern β€” NLP
Extracted critical information from disaster tweets β€” compressed 3,000 tweets β†’ 40 most-critical using LLMs + integer linear programming. Also built a flood-detection pipeline with segmentation for improved classification.
Jul – AugΒ 2023
Hyderabad
DRDO Β· RCI Β· Machine Learning Intern
Log analysis & anomaly detection across 1M+ logs with One-Class SVM, K-Means, Isolation Forests β€” reduced outliers to the hundreds. Used Splunk + Kafka. PoC presented to senior scientists for production exploration.
May – JulΒ 2023
Mumbai
Jio Platforms Limited Β· AI Intern
"Improving Operations with AI" β€” designed & deployed recommendation algorithms (incl. Apriori). Built a stakeholder-facing web app summarizing impact on operational efficiency.

▍ Publications

# Title Venue / Source Year
β˜… Compounding Disadvantage: Auditing Intersectional Bias in LLM-Generated Explanations… ACM FAccT 2026 Β· arXiv:2601.14506 2026
2 Traffic Sign Sensing: A Deep Learning Approach for Enhanced Road Safety cited 5Γ— 2024
3 YOLOv8-Based Detection of Fall Armyworm in Field Crops IEEE INDISCON 2024
4 Automated Animal Intrusion Detection: A Deep Learning Approach Int'l Conf. on Information Technology 2024

Full list β†’ Google Scholar


▍ Toolbelt

GenAI Β· LLMs

Core ML Β· frameworks


▍ Beyond the terminal

  β–Έ Organized festivals @ E-Cell IIIT Pune β€”  1,000+ students, 11 countries   (Oct 2023)
  β–Έ Led Knightmares Chess Club @ IIIT Pune β€”  400+ members                    (Jul 2022)
  β–Έ Model United Nations delegate β€”  diplomacy, global issues, too much coffee

"The best way to predict the future is to audit the models shaping it."

Pinned Loading

  1. Traffic-Sign-Sensing Traffic-Sign-Sensing Public

    The "Traffic Sign Sensing Using Object Detection with YOLOv8 and NAS-YOLO" project represents a cutting-edge application of computer vision .Additionally, the project has resulted in the creation o…

    Jupyter Notebook 2

  2. CareerVaani CareerVaani Public

    Forked from KrishY17/CareerVaani

    Career Prediction System

    HTML 1

  3. Animal-Intrusion-Detection Animal-Intrusion-Detection Public

    We address the alarming rise of animal-inflicted harm in residential areas, proposing an image classification-based identification system for enhanced safety. The research highlights the urgent nee…

    Jupyter Notebook