Aniket Gupta

Aniket Gupta

MS Computer Science @ Stanford University · MLE Intern @ Upstart
LLM post-training · Reinforcement learning · Generative model evaluation

About

I'm a master's student in Computer Science at Stanford (GPA 3.95/4.0), where I work on post-training and evaluation of generative models — reward modeling for video generation, RL fine-tuning of LLMs, and building language models from scratch. My current research on automated video preference scoring is under review at NeurIPS 2026.

This summer I'm a machine learning engineering intern at Upstart, improving the accuracy, runtime, and cost of personal-loan underwriting models.

Before Stanford, I spent 2.5 years as a software engineer at Samsung Electronics HQ (Suwon, South Korea) working on touch input kernel drivers and on-device gesture recognition for flagship phones, and a year at fintech startup slice as AVP of Strategy & Operations. I did my B.Tech. in CS at IIT Kanpur, graduating with Distinction with a 9.94/10 GPA and the Academic Excellence Award in all four years.

Research & Projects

Recreating Video Arenas via Automated Preference ScoringUnder review · NeurIPS 2026

Stanford · 2025–26 · Paper website

Introduced Video Preference Score (VPS), a fully automated preference model that recreates large-scale Elo-based video-generation arenas without human raters. Trained on a newly collected dataset augmenting generated videos with retrieved real-world footage, VPS correlates highly with human consensus and generalizes zero-shot to unseen models and benchmarks. Used as a reward model with Flow-GRPO, it improves WAN-2.1's generation quality by +51 Elo.

Difficulty-Aware Curriculum Learning for RL Fine-Tuning of LLMs

CS 224R (Deep RL, Prof. Chelsea Finn) · Stanford · 2026 · Poster

Studied whether curriculum learning helps LLM post-training across SFT, IPO, and RLOO on arithmetic-reasoning tasks, using an exact difficulty signal computed via dynamic programming over ~490k prompts. Found adaptive (competence-based) curricula lift SFT by +7.1 pass@1 while offline objectives are largely order-invariant — and showed why RLOO's advantage collapses to zero under difficulty re-weighting.

Casting the Graph: Predicting Stars Before They Shine

CS 224W (ML with Graphs, Prof. Jure Leskovec) · Stanford · 2025 · Blog post

Modeled the film industry as a heterogeneous graph — ~37k movies, ~163k people, ~400k edges, with BERT plot embeddings and Laplacian positional encodings — and trained R-GCN and HGT models for cast recommendation as link prediction (HGT AUC 0.76 vs 0.52 ALS baseline) and IMDb rating prediction as node regression (RMSE 0.81 vs 1.04 Node2Vec + XGBoost).

PEFT for GPT-2: Paraphrase Detection & Sonnet Generation

CS 224N (NLP with Deep Learning) · Stanford · 2026 · Poster

Implemented GPT-2 from scratch, then studied how far parameter-efficient fine-tuning can go: LoRA/DoRA on KQV+MLP recovers nearly all of full fine-tuning's paraphrase accuracy (0.882 vs 0.898) with ~7% of trainable parameters, and repetition-aware decoding was the single biggest win for sonnet generation (+6 chrF). Also completed CS 336 (Language Modeling from Scratch, Profs. Percy Liang & Tatsunori Hashimoto), building the full LLM training stack end-to-end.

Extending GemOS

Operating Systems (CS 330) · IIT Kanpur · 2019

Implemented core OS primitives in a teaching kernel: locks, fork/vfork, mmap, and file-system read/write/lseek operations.

Experience

Jun 2026 – Sep 2026

Upstart — Machine Learning Engineer Intern

Remote, USA

Improving the accuracy, runtime, and cost of Upstart's personal-loan underwriting models.

Apr 2024 – Jun 2025

slice — AVP, Strategy & Operations

Bengaluru, India

  • Increased user onboarding conversion by 2% by eliminating friction points across the onboarding journey.
  • Cut Video-KYC turnaround by 30 seconds through streamlined verification and automation.
  • Hardened the user-deduplication system against bypass while surfacing historic duplicates.
Sep 2021 – Mar 2024

Samsung Electronics HQ — Software Engineer

Suwon, South Korea

  • Owned integration and maintenance of touch-input kernel drivers across Samsung flagship and non-flagship devices.
  • Restructured the Android service layer to process input matrices from multiple touch panels, enabling gesture-recognition models on dual-screen devices.
  • Designed linear-time algorithms for touch-gesture recognition, pipelined from kernel to Android service layer.
  • Integrated AI-based recognition models into the Android service layer for flagship devices.
May 2020 – Jun 2020

Samsung Electronics HQ — Software Engineering Intern

Suwon, South Korea

Proposed audio accessibility features for Android phones and verified their patentability, outlining control/data flow and technical constraints.

Jun 2019 – Jul 2019

University of Washington — Research Intern

Seattle, WA · with Prof. Sreeram Kannan

Built a Rust-based blockchain application to empirically test research on increasing transaction throughput and reducing confirmation latency.

Education

2025 – 2027

Stanford University — MS, Computer Science

GPA: 3.95/4.0

CS 229 · Machine Learning CS 224N · NLP with Deep Learning CS 224R · Deep Reinforcement Learning CS 336 · Language Modeling from Scratch CS 224W · ML with Graphs CS 227B · General Game Playing ENGR 245 · Lean LaunchPad
2017 – 2021

IIT Kanpur — B.Tech., Computer Science & Engineering

GPA: 9.94/10 · Graduated with Distinction

Mathematics for CS I–III Advanced Algorithms Algorithmic Game Theory Programming for Performance Parallel Computing Operating Systems Compiler Design Topics in Topology Mathematical Economics

Awards & Honors

Teaching