Backend & Distributed Systems
Spring Boot services, Kafka streams and Elasticsearch clusters built for throughput — 100K+ events per minute, 40% lower query latency, dual-write consistency with zero data loss.
Winner
Mastercard Innovation Challenge @ GFF 2026
Mastercard GFF 2026
Open to SDE, Data, ML & AI Engineering roles · 2027
Aditya Rai
I build
Final-year CSE undergrad at IIIT-Delhi, working where backend systems meet data at scale — enterprise data graphs, streaming pipelines that clear 100K events/min, and agentic AI that has to be auditable, not just impressive.
- 3Engineering internships
- 100K+Events/min in production
- 3×National hackathon podiums
- 400+DSA problems solved
01 — About
Engineer first, curious always
I'm a Computer Science undergrad at IIIT-Delhi who gravitated toward the unglamorous half of software — the part where data has to arrive on time, stay correct, and explain itself afterwards.
That pull has taken me through three internships: modelling an enterprise data graph and building AWS Glue/Athena pipelines at TBO.com, running a real-time analytics platform on Spring Boot, Kafka and Elasticsearch at LogiCoy, and shipping an ML revenue-forecasting system at HCLTech. The through-line is the same: make the pipeline dependable, then make it fast.
Outside of coursework I'm usually mid-hackathon. Most recently I led my team to first place at the Mastercard Innovation Challenge at Global Fintech Fest 2026 with Adversarial Payment Fraud AI, out of 1,024+ submissions. Drishti and TBO Assemble came from the same habit — projects built under a clock, which turns out to be a good forcing function for deciding what actually matters. I like teams that argue well, ship often, and stay humble about it.
Data Engineering
AWS Glue ETL over S3 data lakes, tuned Athena queries, Redshift analytics models and ontology-driven data graphs — with the quality validations that keep the numbers trustworthy.
Applied AI
Fraud-detection ensembles, LightGBM risk models, CNN classifiers and LLM document intelligence — shipped with deterministic policy layers and tamper-evident audit trails, at sub-10 ms inference.
02 — Experience
Where I've shipped
-
May 2026 – Jul 2026
Gurgaon
Software Engineering Intern
TBO.com
- Contributed to an Enterprise Data Graph platform through ontology-driven entity modelling, relationship analysis and AI-enabled business intelligence workflows.
- Developed AWS Glue ETL jobs and optimised Athena queries over Amazon S3 data lakes for scalable KPI computation and analytics workloads.
- Built KPI computation frameworks, data-quality validations and Redshift analytics models supporting enterprise-scale reporting.
-
Aug 2025 – Oct 2025
Remote
SDE Intern
LogiCoy
- Spearheaded a real-time user analytics system on Java Spring Boot, Elasticsearch and Apache Kafka, processing over 100,000 events per minute.
- Refined Elasticsearch indexing strategies and integrated advanced filtering, cutting analytics query latency by 40%.
- Implemented Release History Management REST APIs with dual-write sync between MySQL and Elasticsearch — zero data loss across 50K+ versioned records, covered by JUnit.
-
May 2025 – Jul 2025
Noida
Summer Intern
HCLTech
- Architected an end-to-end ML-based Revenue Forecasting Platform using Python, Scikit-learn, SQL and automated ETL pipelines — forecasting accuracy up 15%, manual effort down 90%.
- Integrated interactive business-intelligence dashboards in React and JavaScript, enabling drill-down analytics and improving insight generation by 25%.
03 — Projects
Things I've built
Three flagships up front — led by the Mastercard-winning build — then the ones that taught me the fundamentals.
1st · Mastercard GFF 2026
Aug – Sep 2026
Adversarial Payment Fraud AI
Red-teaming AI for GenAI-era payment fraud
A platform that attacks payment systems the way GenAI-era fraudsters would, then learns to stop them. It simulates 67 attack vectors across 9 fraud families and 3 payment data planes, and defends with a stacked detection ensemble — gradient boosting, random forest and isolation forest over behavioural features, backed by security guards. It took first place at the Mastercard Innovation Challenge @ GFF 2026.
- 67attack vectors simulated
- 71.6%recall at 0.535% false-positive rate
- 9 × 3fraud families × payment data planes
Flagship
May – Jun 2026
Drishti — Agentic AI Loan Officer
An agentic underwriting platform that collapses a 40+ field loan form into a five-minute voice conversation. Behind the voice sits a deterministic credit-policy engine, LightGBM risk scoring, three-stage fraud detection and a tamper-evident SHA-256 audit chain — so every decision can be replayed and defended.
- <10 msmodel inference
- 1.21 smedian response
- 40+ → 0form fields
Flagship
Feb – Mar 2026
TBO Assemble — AI Group Travel Platform
A SaaS platform that reads hotel contracts so humans don't have to. OCR and PDF parsing feed GPT-4o document intelligence, which turns unstructured contracts into structured, queryable data — wired into a backend built for real operational load.
- 42+REST APIs
- 13-modelrelational schema
- Real-timesync + analytics
AI-Generated Image Detection
Binary classifier separating real photographs from synthetic ones. A custom CNN reached 84.76% accuracy; classical SVM and Random Forest baselines over HOG, LBP and SIFT features held 78%.
RISC-V Assembler & Simulator
A full assembler and cycle-simulator for the RISC-V ISA, written from scratch in Python — arithmetic, logical and floating-point instruction support, with register and memory state traced per cycle.
Angry Birds Clone
A physics-driven game clone in Java built on OOP principles, using LibGDX for rendering and Box2D for collision and trajectory simulation — with procedurally generated infinite levels.
SahaYan
A donation platform bridging donors and NGOs for non-monetary giving — routing unused goods to the people who need them. Designed end-to-end in a team of six.
Nothing in this category yet.
04 — Skills
The toolkit
Languages
Frameworks & Libraries
Databases
Cloud & DevOps
Data & Architecture
05 — Honors
Recognition
Three national hackathon podiums — most recently, first place at Mastercard's Innovation Challenge.
1st Place
Mastercard · Sep 2026
Mastercard Innovation Challenge @ Global Fintech Fest 2026
First place in AI Defence Lab for Payment Security, ahead of 1,024+ submissions nationwide. I led our team through repeated rebuilds, taking Adversarial Payment Fraud AI — a red-teaming platform for GenAI-era payment fraud — from an idea to the winning stage.
- #1of 1,024+ submissions
- Team leadidea to winning build
- AI Defence Labpayment-security track
VOYAGEHACK 3.0 — First Runner-Up
TBO.com · ranked top 2 of 11,000+ participants.
TenzorX 2026 National AI Hackathon — Second Runner-Up
Poonawalla Fincorp · ranked top 3 of 27,000+ nationwide.
Goldman Sachs India Hackathon 2025 — Rank 1
Achieved Rank 1 and selected as a Semi-finalist.
Amazon ML Summer School 2025
Selected for the programme led by Amazon Scientists on advanced ML topics.
JEE Advanced AIR 6841
99.45 percentile in JEE Mains.
INMO & INAO
Selected for the Indian National Mathematical Olympiad and the Astronomy Olympiad.
06 — Contact
Let's build something
Roles, collaborations, or a good technical argument — my inbox is open.