Raza Hashmi

I build products and run ML research—especially on uncertainty, scaling and interpretability. Over the last decade I’ve taken products from fuzzy problem statements to shipped systems—connecting user needs, UX, data/metrics, and implementation to build solutions that are privacy-conscious, decision-useful, and scalable.

On the research side, I work on uncertainty, interpretability, and the learning dynamics that shape model behavior.

Now

• Drafting papers (in progress; not yet submitted):
• Pathway-Enhanced Uncertainty Estimation — uncertainty from internal pathway/prototype consistency
• Diffusion-based Uncertainty — sampling-driven uncertainty with theory checks and baseline comparisons
• Difficulty Is Relative — a unified suite to measure and predict class-pair difficulty
• Building evaluation tooling for distribution shift, calibration, and rejection/coverage tradeoffs

What I’m known for

• Turning messy, ambiguous problems into clear product strategy, experiments, and shipped outcomes
• Designing incentive-aligned systems (metrics, scoring, dashboards) that people actually trust and use
• Building privacy-preserving approaches that don’t compromise analytical usefulness
• Partnering effectively across engineering, design, and stakeholders with crisp communication

Highlights

Product leadership (SHAPE — Survey & HR Intelligence Platform)

As Head of Product, I helped evolve SHAPE from an advanced survey product into an HR intelligence platform with:

• Org hierarchy mapping for complex, multi-layered organizations
• Role-based access and reporting for employees, managers, and executives
• Multi-level reporting architecture (aggregation/disaggregation across org structures)
• Proprietary scoring approaches for nuanced survey analytics

Selected scale / proof points:

• 12 measurement dimensions (“Explorers”) across You / Team / Org
• Validated across thousands of employees, industries, and regions
• Instant personal report with 60+ report cards per employee

Selected artifacts to showcase some of the work that i did:

• Privacy-centric algorithms demo:
• Scoring system Whitepaper:

Click here for more work details

Building measurement frameworks (World Flourishing Organization)

As a founding member at World Flourishing Organization, I helped develop a framework to identify and measure flourishing in organizations—bridging research concepts, practical measurement, and what it would take to run this at scale.

• The Framework i designed

ML Research

• Presented: “Explainable AI: Object Recognition With Help From Background” (ICLR 2022, CSS Workshop)
• Current work (drafts): pathway-based uncertainty, diffusion-based uncertainty, and difficulty/learning-dynamics methods (see “Now” above).

Interests

• Tech startups building creative solutions to real problems
• Products where measurement, incentives, and decision-making under uncertainty matter
• AI-enabled products where model behavior, evaluation, and deployment tradeoffs matter
• Research collaborations in uncertainty, generalization, and interpretability

Writing & side projects

• Reinforcement learning project: “Can an AI Learn the Art of Valet Parking?”
• Essay on using GenAI responsibly and staying in the driver’s seat
• An iterative trolley-problem experiment + survey

Browse all posts: /year-archive/