Kartik Bathla

Identified that quant researchers lose time re-verifying AI-generated notebooks; defined OpQuant as a persistent, AI-native research workspace and took it to a live private beta with users from Goldman Sachs, Millennium and Squarepoint. Built a custom agent harness, a persisted tool-calling loop across 22 tools, with resumable state, per-step traceability, and approval gates.

Created ML based mid frequency equity alphas balancing signal strength, turnover, drawdown and consistency across market regimes. Built a hybrid search system (vector + BM25) for a data catalog with 47k+ data points, significantly improving datapoint discovery in research workflows.

Built a deep learning pipeline to automate the detection of anatomical landmark points in 3D point cloud intraoral scans using Point Transformer v3 architecture based models. Achieved 0.79 mAP / 0.84 mAR at 2 mm tolerance under the MICCAI 3DTeethLand'24 protocol, ranking fifth globally, with 0.9-second inference per scan.

Worked across Quant-Strategies, Oil & Gas Options, and Global Macro desks conducting research and providing live trade support. Conceptualised and drove a geopolitical intelligence system from 0 to 1 for the options desk, which predicted US retaliation ahead of B-2 bomber strikes in June 2025.

Owned development and production operation of ML/data systems supporting J.P. Morgan recommendation engines, client-engagement and personalization products. Took a client-profiling ML system from research to production, engineering portfolio and behavioural features, and operationalised daily inference, monthly retraining and weekly model-drift evaluation.

Team Scrum Master - ran sprint planning and prioritisation across engineering and business stakeholders and reported delivery progress to senior leadership. Built and maintained a reusable 120+ feature/marker layer for downstream recommendation and client-engagement applications, integrating trading, pricing and enterprise data sources.

Founded and led a 10 person team building an e-receipts platform, owning product direction, technical architecture and execution. Took the product from concept through working prototype while balancing feasibility and commercial considerations.

Built failure analysis tooling, prototyped for database switch, designed FIX-message indexing methods that reduced diagnostic search P90 time from 420 secs to 2 secs.

Pursued a Master of Technology in Artificial Intelligence at the National University of Singapore from August 2024 to October 2025.

Activities and societies: Merit based Tuition Scholarship | 1st Place, HKUST Robot Design Competition | Exchange Semester, UMass Amherst

Earned a Bachelor of Engineering in Computer Engineering from the Hong Kong University of Science and Technology from September 2017 to December 2021. Received a merit-based tuition scholarship and placed first in the HKUST Robot Design Competition.