I work across product, program, GTM systems, enterprise transformation and responsible AI.
My focus is usually the space between strategy and execution: taking an ambiguous business problem, turning it into a product, system or operating model, and bringing the people, technology, governance and measurement together to make it work.
Product strategy, enterprise platforms, CRM, HCM, workflow automation and AI enabled business systems.
I am particularly interested in products where technology has to fit into real operating processes rather than exist as a standalone feature.
Cross functional programs involving product, engineering, operations, commercial teams, services, legal, finance and customer facing organizations.
I work on turning large initiatives into clear workstreams, dependencies, decisions, readiness measures and measurable outcomes.
Launch planning, commercial readiness, enablement, adoption, operating processes and the systems behind them.
I am interested in how product, CRM, automation, analytics and AI can improve the path from product strategy to customer adoption.
Practical controls for AI systems including evaluation, runtime safety, human review, governance evidence and agent containment.
Exploring how model selection, routing and infrastructure decisions can reduce inference impact while preserving useful system behaviour.
Exploring how LLM requests can be routed using task complexity, model capability, latency requirements and carbon intensity.
CAIR combines model routing, carbon signals and audit evidence to examine whether inference can become more efficient without treating every request the same way.
Repository: sa1-carbon-inference-router
A practical AI safety project exploring how indirect instructions can redirect agent behaviour and how detection, classification and containment controls can respond.
Repository: ac4-agent-hijacking
Exploring how recurring AI evaluations can become operational governance workflows involving monitoring, escalation, human review and auditable evidence.
Repository: ai-compliance-agent
An evaluation framework examining whether AI safety responses remain consistent across demographic and contextual variations.
Repository: ai-bias-evaluation-framework
A verified longitudinal dataset supporting Women's Safety AI research, including source provenance, verification records, analysis code and reproducible datasets.
Repository: india-rape-statistics-ncrb
Carbon Aware Inference Routing for Large Language Models
Research into carbon aware model routing, inference efficiency and auditable sustainability signals.
Women's Safety AI Evaluation
Research into demographic invariance, safety response quality and governance evidence for AI systems.
I enjoy connecting areas that are often treated separately:
Business problem → Product thinking → Systems design → Program execution → GTM readiness → Adoption → Measurement
And increasingly:
AI capability → Evaluation → Governance → Runtime controls → Evidence
That intersection is where most of my projects live.
I am using GitHub increasingly as a working environment rather than only a publishing destination.
Current areas of practice include pull requests, code review, CI workflows, issue driven development and collaborative open source contribution.
Interested in collaborating on practical work involving enterprise AI, product systems, AI governance, safety, evaluation and sustainable AI.
🌐 Portfolio: https://preetibuilds-33d6f6da.vercel.app/