Building AI systems that work at real-world scale. Currently at Blinkit, previously Scopely.
I care about the unglamorous middle — between "demo works" and "holds up at 3 a.m."
At Blinkit: Rancher, Kubernetes, Go. At Scopely: production LLM tooling, eval pipelines, architectures that don't fold under a weird prompt.
Before LLMs, cybersecurity at CDAC — Flask, ping floods, iptables. Research at CIMFR on social-media sentiment. Published in IEEE, co-authored a CRC Press chapter on DDoS.
Cybercrime complaint classifier. Ensemble of RF, SVM, Voting, TextCNN across fraud, phishing, identity theft, ransomware.
Time-of-use tariff optimizer for solar + grid. Shifts load by price signal.
AI research assistant on GPT-5.
Socratic DSA tutor. Won't hand you the answer.
Leave-Mgmt + Google MCP servers for Claude.
NL → cloud diagrams + Terraform.
Network auditing in a Flask app.
Real-time monitoring, thousands of events/month.
An ML/DL ensemble (Voting Classifier + TextCNN) that classifies cybercrime complaints into fraud, phishing, identity theft and ransomware. Evaluated on real complaint data; written with an intentional focus on where the model is still wrong, because that's where the interesting work is.
A survey + proposal chapter on how modern threat-intel feeds can be folded into DDoS prediction pipelines for cloud-native workloads.
From day one, she brought genuine curiosity and a strong willingness to learn. She could grasp new concepts and apply them with minimal hand-holding. I'd enthusiastically recommend her to any team looking for a driven, passionate engineer.
I'm graduating in September 2026 and actively looking for AI Engineering, ML Research and SWE roles. If you're building anything in the unglamorous middle — LLM infra, evaluation, reliability, ML ops — I'd love to hear from you.
vaibhaveesingh89@gmail.com ↗