Patent-intelligence venture
Praviar
In-production pharmaceutical patent-intelligence venture, going to market, with fixed checks, bounded legal reasoning support, and private implementation detail.
Complete technical profile
Peter McCann Strain builds the whole system: governed AI platforms, private assistant infrastructure, patent-intelligence products, real-time voice operations, evaluation harnesses, production workflows, and research methods.
Patent-intelligence venture
In-production pharmaceutical patent-intelligence venture, going to market, with fixed checks, bounded legal reasoning support, and private implementation detail.
Voice-operations venture
In-production AI receptionist venture going to market, focused on missed-call recovery, booking support, privacy, and launch-gated operations.
Enterprise agentic AI at SThree
Multi-agent orchestration, synthetic candidate systems, evaluator architecture, self-improvement research, reviewable state, fairness diagnostics, red-team layers, audit records, and production control.
8 agents · private daily use
An ongoing independent private assistant with durable memory, proactive background work, and quality monitoring.
0 mismatches · 50 runs
Co-founded Datamise; built claims infrastructure and a valuation engine that matched a separate calculator penny-for-penny.
P0–P12 patterns
Benchmark study of around 13 orchestration patterns with judge-reliability panels, citation audits, and equivalence testing.
180× cheaper
CV skill tagging rebuilt around retrieval, filters, calibration, and a 7 kB head — zero LLM calls in the production hot path.
Oxford · Astroscale · CivTech
An 89%-accurate medical-imaging classifier, satellite sim-to-real vision, and public-sector AI strategy across hard domains.
The throughline is range under constraint: regulated hiring workflows, reviewable legal calculations, satellite computer vision, and Oxford medical-imaging research all demanded traceability rather than a demo.
The ideal role is senior or staff-level AI engineering where architecture, evaluation, governance, and product delivery all matter.
Continue exploring
Agentic systems, bounded public detail, reviewable state, and production evaluation.
OpenBias testing, adversarial datasets, and auditable fairness metrics.
OpenDeep-research benchmarks, judge panels, citation audits, and equivalence tests.
OpenProduction AI economics, hot-path removal, retrieval filters, and tiny heads.
OpenAn ongoing independent case study on private assistant memory, proactive operation, and quality monitoring.
OpenCurrent founder venture in bounded pharmaceutical patent intelligence, with protected product mechanics kept private.
OpenCurrent founder venture in real-time AI voice operations for dental practices, with demo video coming soon.
OpenThe longer professional narrative and contact route.
Open