Sakshat Sachdeva

01 — What I do

Where I move the needle.

Building, nerding out and gyming are three things I enjoy the most. I write low level code, or solve puzzles in my free time. Shipping apps with 5K+ downloads, co-authoring a paper under review at ICLR 2027, winning national competitions across tech, consulting and MUNs, teaching quantitative strategies, or being invited as a guest to universities like DU — often in the same semester. The thread: Time management and competitive spirit wins. Kaizen.

5K+

Play Store downloads for Pocket-Pixel — 60 FPS Game Boy emulator, C++ core with JNI & WASM builds.

96K+

Teams outperformed across national competitions — 1st at AWS AI for Bharat, 3rd at NMIMS, Top 150 Goldman Sachs.

8.53 cgpa

Top 5 in stream at IIIT Delhi. Dean's List 2024. District Topper — 99th percentile in CUET & Boards.

How I move the needle

Build and shipGame Boy emulator (5K+ downloads), lock-free HFT engine (<120ns p50), interpreter with self-written GC, production agentic AI assistant shipped at GEP. End-to-end, every time.
Win under pressure1st at AWS AI for Bharat (96K teams), 3rd at NMIMS (1000+ teams), Top 150 Goldman Sachs Hackathon, Runner-Up Contingent at IIT Mandi (4 wins). Timed constraints bring out my best work.
Teach and leadPresident of the Placement Committee and Quant Society at IIIT Delhi, TA for Advanced Deep Learning, Guest Speaker & Judge at DU colleges, ex-Student Senate. Leading from the front, always.
Follow the moneyNSE micro-tick pipelines, EMA signal generation, Monte Carlo American options pricing, Black-Scholes model from scratch. Market microstructure is measurable — I measure it.
Ask why, not whatCausal inference on time-series foundation models at IBM Research, DiD and synthetic control on meme-stock volatility, a KL certificate that predicts estimator bias. Correlation is the question, not the answer.

turning the dial · always building

02 — Work Experience

The road so far.

G

B

P

T

S

B

N

S

G

I

G

Gigtern

CTO & COO · Full-time · Delhi

Nov 2023 – May 2025

Led as CTO & COO — hyper-local gigs and internships platform.

B

Barkedo

Founder's Office · Full-time

Nov 2023 – May 2025

Indie puppy adoption platform. Facilitated 50+ adoptions and led community events.

P

ProCohat

DevOps Intern · Maharashtra

Jun – Aug 2024

Full-stack dev & DevOps at a hybrid studio — end-to-end product work.

T

Tale of Humankind

Program Head · Delhi

Jun – Dec 2024

Ran YCB bootcamp at IIT Delhi — mentors, logistics, execution.

S

Shipd

Problem Solver · Remote

Mar – May 2025

Project Loki — OSS contributions to SciPy & pip before program ended.

B

Banavo.AI

SWE Intern · Full-time

May – Jul 2025

Built LLM integration and RAG pipelines at an early-stage AI startup.

N

NX Block Trades

Quant Researcher · Noida

Aug 2025 – Jan 2026

NSE micro-tick pipelines, EMA signals, and alpha research across regimes.

S

Scrite.io

Open Source Dev Intern · Remote

Jan – May 2026

PDF → Fountain converter in C++ using Qt & Tesseract OCR. World's first.

G

GEP Worldwide

SWE Intern · Platforms (Agentic AI) · Mumbai

May – Jul 2026

Shipped a production agentic AI assistant inside an enterprise Word add-in — intent planning, approve-before-apply edits, secure Azure OpenAI proxy.

I

IBM Research

GRM Intern · BTech Project · Delhi

Jan 2026 – Present ★

Causal inference with time-series foundation models — KL certificate for estimator bias (corr. 0.976), now causal task arithmetic via model merging.

03 — Honours & Awards

Things that stick.

Research

ICLR 2027Co-author, Valence Asymmetries in Language Models: Representation, Behaviour, and Self-Report — under review. Emotion vectors across LLM families organise into a valence–arousal geometry; steering shifts refusal behaviour.
IBM ResearchBTech Project, Jan 2026 – Present — causal inference with time-series foundation models. A KL-divergence certificate predicts estimator bias (corr. 0.976 across 11 representations); frozen vs. fine-tuned Chronos-Bolt and TSPulse against 5+ baselines.
Causal · DiDWhen Reddit Broke Wall Street — causal effect of subreddit activity on GME volatility via Difference-in-Differences and Synthetic Control, with event studies.
Crypto · MLNeural Cryptanalysis of Lightweight Ciphers — CNN and MLP distinguishers for LED, LBlock, PRINCE and TEA; comparative review of ML-based cryptanalysis benchmarks.
exSPLORe '26Ashoka University — 1 of 10 undergraduates among 160 researchers for a five-day quantitative ML research school with scientists from Google DeepMind, MSR and Stanford.

Achievements

1st / 96,000AI for Bharat, AWS (Amazon) — first place out of 96,000 teams in a nationwide AI competition.
FeaturedIndia Today — featured for winning AWS AI for Bharat, national media coverage.
Dean's ListDean's List Excellence Award, IIIT Delhi — recognised for academic excellence, 2024.
3rd / 1,000+NMIMS National Hackathon — third place out of 1,000+ teams across the nation.
Top 10 / 600+IIM Calcutta National Case Competition — national finalist at India's premier B-school.
Runner-UpIIT Mandi — won 4 competitions; IIIT Delhi secured 2nd position overall as contingent.
Top 150Goldman Sachs Hackathon — across IITs, IIITs, and NITs nationally.
Alumni AwardMaxfort School, Rohini — School Alumni Award for making strides in the field of technology.

Speaking & Mentoring

Speaker & JudgeSri Venkateswara College (DU) — delivered talks and evaluated premier competitions at University of Delhi.
Speaker & JudgeShaheed Bhagat Singh College (DU) — invited at University of Delhi business event.
Speaker & JudgeIIIT Delhi — invited to judge Nok Jhok, IIIT Delhi's premier debate tournament.
Mentor AwardSCIF × IIT Delhi — mentored national AI hackathon at IIT Delhi; received award for leadership.
Mentor AwardTale of Humankind — selected to mentor India's leading 40 high school students.

Leadership

PresidentPlacement Committee, IIIT Delhi — 2026 – Present; Core Member, May 2025 – 2026.
PresidentQuant Society, IIIT Delhi — 2026 – Present. Previously taught Black-Scholes and quantitative strategies to the batch.
TAAdvanced Deep Learning, IIIT Delhi — Teaching Assistant, Monsoon 2026.
Co-FounderGigtern — hyperlocal gig platform for students, scaled to ₹30L+ ARR; backed by Shiv Nadar University, VIT Arunachal Pradesh and IIIT Delhi.
SenateBatch Representative, Student Senate, IIIT Delhi — Aug 2024 – Jul 2025. Founder, Undergraduate Research Club. Lead, Debate Society.

Invited Events

Sole Student RepUS–India AI Workshop — sole student representative at high-level bilateral policy and research summit with faculty and professionals.
GoogleGoogle Student Launchpad — selected for Google's flagship innovation and software development incubator for handpicked students across the country.
McKinseyMcKinsey Forward Program — chosen for selective leadership and problem-solving cohort for high-potential undergraduates.
1 of 10exSPLORe 2026, Ashoka University — one of 10 undergraduates among 160 researchers from TIFR, IISc, IIT Bombay, ISI Kolkata; scientists from Google DeepMind, MSR, and Stanford.
YC IndiaY Combinator Startup School India — Bangalore cohort; direct engagement with YC partners and leading Indian tech founders.
1 of 30Grad Capital's Lead 30 Meetup, Delhi — invited to meet face-to-face with the founders of India's biggest accelerator.

Coursework

CSE641Advanced Deep Learning — as Teaching Assistant
CSE511Computer Architecture
CSE643Artificial Intelligence
CSE583Software Development using Open Source
CSE544Computer Vision
ECO523Causal Inference
CSE342Statistical Machine Learning
CSE231Operating Systems
ECO221Econometrics
CSE202Fundamentals of Database Systems
CSE201Advanced Programming

04 — How I found my way in

A short story, roughly chronological.

Origins

I was a commerce kid in high school, mathematics and economics, not the typical CS pipeline. Two seniors at IIIT Delhi told me about the boards and olympiad route in, and that changed everything. Once I got there, I fell into algorithms, puzzles and low-level systems almost by accident.

The people I met didn't help. They had these grand, specific visions for what they wanted to build, and that kind of energy rubs off on you. It pushed me into some of the most amazing programs, research experiences and opportunities I could have asked for. Truly grateful.

Projects

Built Pocket Pixel, a GameBoy emulator for Android with a C++ core running through JNI, hitting 60 FPS across 100+ titles. It sits on the Play Store with 5000 downloads and a 4.5 rating. Then HFT-Core: a lock-free SPSC queue, cache-aligned order book, AVX2 price search and an RDTSC profiler reporting P50 to P999. Also IIIT Lang, an interpreter for IIIT Delhi slang with a self-written garbage collector, God's Eye, real-time visual SLAM on a 27g nano-drone, and an Angry Birds clone in Java that I probably spent too long on.

The throughline is that I like building things that have real mechanics underneath: emulators, interpreters, order books, SLAM pipelines. Stuff where you can't fake the internals.

What I'm working on

At IBM Research: causal task arithmetic, treating fine-tuning sets as intervention arms so the interference term of merged task vectors identifies a causal interaction effect. In Computer Architecture: a sparse-aware RISC-V vector study on Spike, 270 SpMV runs showing lane utilisation collapse to 12% at 99% sparsity.

On the side, hft-core keeps getting faster, plus stochastic calculus, market microstructure and the C++ memory model. Always looking for the next thing that bends my brain.

↳ IIIT Delhi, BTech CS · graduating July 2027 · open to 2027 systems & quant roles