Research
Four areas, each organised around questions that are still open. Topic pages separate established background, my own results and what remains unresolved, and link to the papers and artefacts behind every claim.
Agent Evaluation and Coordination Research
Escalation, refusal to fabricate, multi-agent coordination, judge robustness and ranking stability, with papers, artefacts and limits.
Topic
Multi-Agent Coordination and Conflicting Work
Duplicate effort, conflicting edits and lost writes between coding agents, measured from a coordination log rather than from pull requests.
Topic
RAG Knowledge Conflicts and Input Regimes
How conflict detectors for RAG behave across input regimes: five findings from an audit, an input-regime ladder and open questions.
Generated-Code and GPU-Kernel Correctness Research
Correctness oracles, test-input generation, calibrated tolerances and a 26-op failure corpus for LLM-generated kernels.
Topic
Efficient Inference and Distillation Evaluation
Open research direction: matched protocols, resource accounting and task-specific degradation for compressed and distilled models.
Commissioned study
Generated GPU Kernel Correctness Studies
Test generated GPU kernels with operator-aware oracles, calibrated tolerances and adversarial inputs.
Federated Learning and Privacy Research
Federated learning under imbalanced data, and the evidence a distributed-ML privacy claim needs: baselines, metrics, threat models.
Decentralised Systems and Protocol Research
DePIN incentives, composability and MEV: protocol research with stated assumptions, models, experiments and limits.
Topic
DePIN Mechanism and Incentive Evaluation
Test a DePIN incentive mechanism with stated assumptions, an agent model, sensitivity analysis and a reproducible simulation.
Topic
Composability, MEV and Protocol Mechanisms
Formalise the assumptions behind fairness, MEV and cross-rollup atomicity claims, then search for counterexamples.
Looking for practitioner-oriented write-ups rather than papers? See the technical guides on governed agents, agent reliability, code correctness, local inference, robotics and blockchain agent security.
Elsewhere
dipankar.co
AI enablement, contract AI engineering and AI delivery — when you need something built, fixed, operated or adopted.
dipankar.name
Hands-on technology leadership mandates — when you need someone to own a technology function with organisational authority.
dipankar.org
Board advisory, executive workshops, institutional programmes and speaking — when the output is a leadership decision, a programme or a talk.