Software Engineer @ Winzo Games | ML Systems & Backend Infra | C++, Distributed Systems, Low-Latency, CyberSecurity | Codeforces Expert | IIT Madras CSE’25

Delhi, India

Delhi, India





BTech, Computer Science and Engineering
Indian Institute of Technology, Madras

High School
Maharishi Vidya Mandir Senior Secondary School
Join Sakthi and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal BossSoftware Engineer @ Winzo Games | ML Systems & Backend Infra | C++, Distributed Systems, Low-Latency, CyberSecurity | Codeforces Expert | IIT Madras CSE’25
Machine Learning Software engineer with a systems and cybersecurity background, currently building distributed backend infrastructure and data science systems at Winzo Games (100k+ DAU) — from low-latency Kafka/Redis pipelines to ML-driven fraud detection and churn prediction. Codeforces Expert (1712, 600+ problems) with C/C++ roots in offensive security: placed 5th globally at eCTF 2024 (200+ teams, ahead of CMU, MIT) Open to backend and C++ engineering roles where systems depth and performance matter. B.Tech CSE, IIT Madras 2025 | AIR 645, JEE Advanced 2021
Chennai
claude code
claude
github copilot
obsidian
Data Science and ML Software Engineer at WinZO, Jun 2025 - Present · 1 yr 3 mos, Delhi, India, - Architected a distributed, event-driven backend (Kafka, Celery, Redis) powering real-time, multi-modal interactions for 100k+ daily active users, with observability via OpenTelemetry and schema validation via Pydantic - Re-architected compute-heavy Spark/Airflow pipelines, cutting compute cost 65% on the platform's top 10 highest-load jobs by eliminating redundant full table scans - Built ML-driven fraud detection — analyzing 200k+ poker games to improve collusion detection accuracy by 40% - Built an NLP-to-SQL pipeline on AWS Bedrock with RAG-based retrieval, accelerating analyst query time 20x (60 min → under 3 min)
Data Science Intern at WinZO, May 2024 - Jul 2024 · 3 mos, Delhi, India, - Developed a novel device fingerprinting algorithm to uniquely identify and mitigate referral and collusion fraud, achieving 99.8% accuracy across a user base of 175 million+ using Gradient Boosting models (LightGBM) - Ideated and built a custom web scraper on Android documentation to collect GDPR-compliant device-specific identifiers, feeding the fingerprinting pipeline
BTech, Computer Science and Engineering, Indian Institute of Technology, Madras
High School, Maharishi Vidya Mandir Senior Secondary School