SUMIT
01HOME02ABOUT03APPROACH04PROJECTS
BOOK A CALL

Selected Work. Built For Humans.

Explore AI platforms and agentic experiences.

VIEW WORK>
SUMIT KUMAR9YEARS IN DESIGN

AI Design.
Human Trust.

Senior Product Designer specialising in agentic systems, enterprise AI, and human-in-the-loop experiences. 9 years turning complex AI into products people actually trust.

View Work
sumit.uxai@gmail.com
BUILDING WITHClaude CodeCursorFigma
+ x X 8 0 @# % $ S
# # @ @0 0 8 X
+ + x xX X 8 0
8 0 S S@ # % $
+ + x x 80 S S #

SCROLL TO CONTINUE

My Design Process

LAYER 4- Working Product
LAYER 3- Interaction Design
LAYER 2- Product Strategy
LAYER 1- Research

Research. Design. Build.

sumit.uxai@gmail.com
SUMIT 9 years
Agentic systems, enterprise AI, human trust.
research    people · needs, context, constraints
strategy mapped · confidence, handoffs, control
interaction
prototype    discover frame map prototype test
trust   discover map frame prototype test ship
testing   review approve explain correct
evidence  signals evidence confidence limits
build
VIEWExplore the live product
Research → design → working software.
{  DESIGN PRINCIPLE  }
CLARITY
MAKE AI UNDERSTANDABLE
$ SENSE — start with the problem
# Research, stakeholder alignment, edge cases
$ INTERPRET — map trust and autonomy
# Decide what AI can do, and when humans confirm
$ GENERATE — test competing directions
# Challenge tradeoffs before drawing the screen
$ NAVIGATE — design working experiences
MAP · SESSIONS · ARTIFACTS · MEMORY
$ ASSEMBLE — build and ship
LIVE PRODUCT: AIDEX
Review flows, confidence states, and source traceability.
$ LEARN — check, correct, improve
# Use evidence; label expected outcomes honestly
01

Start with the problem

Research, stakeholder alignment, and edge-case mapping before any wireframe.

02

Design the whole system

Map agent behaviour, confidence states, and human handoff moments before the interface.

03

Build working products

Research, decide, design, build, ship, and improve — as demonstrated in Aidex.

What I Bring
To Every Project.

Explainable Matching

HireU shows why a role matches, and what the profile does not cover.

MATCH REASONING92%profile coverage

Evidence you can check

Aidex links each claim to its exact source and shows what was searched, skipped, or unreadable.

verified_answer.json
source sentenceMSA-2023 §11.4
checkTRACE source
checks passed7/7
actionsASK → TRACE → APPROVE

Design to working product

I work embedded with AI engineers and data scientists, from research and system behaviour through to a live product.

One guided workflow

HireU connects Find, Match, Prepare, Review, Apply and Track in one reviewable journey.

approval-gatesmatch-reasoningprofile-contextrésumé-versioncover-lettersaved-searchreview_changesapprove_send

Interaction & Prototype Design

Interactive prototypes make complex AI behaviour tangible, from confidence states to human handoffs.

Prototype states
UX flow exploration
Flow states
R1R2R3R4R5R6R7R8R9R10R11R12

Visible AI changes

HireU shows what changed, why it changed, and what is about to be submitted.

candidate_profile.jsonoriginal_profile.json
profile.name-"Sumit"
profile.role-"Senior Product Designer"
profile.experience-"9+ years"
review.changes-[ ... visible changes ]
candidate.request-"tailor résumé to this role"
ORIGINAL KEPT VISIBLE
reviewable_changes.jsonprepared_application.json
proposal = {
"instruction" : "tailor résumé to this role",
"visible_changes" : {
"review" : "show résumé diff",
"material" : "résumé experience"
}
}
# Original experience stays intact
# Changes are explained before approval
NOT SENT UNTIL APPROVED

Rules that preserve control

HireU tailoring only adds what is already true of the profile. Final submissions wait for explicit approval.

ASK — ask about the documents
$ ASK — what are our termination obligations?
# TRACE opens the highlighted source sentence
Reviewable MSA-2023 §11.4 · Exhibit D, row 7 [source linked]
$ TRACE — inspect the proof
# Claim, evidence, confidence, limits, next action

AI Design For
Human Trust

I design AI systems that humans can understand, trust, and act on.

Agentic UX Design-

Designed multi-agent systems where AI reasons, recommends, and acts — with humans staying in control at every critical decision point.

AI Transparency+

Made complex AI behaviour legible to non-technical users through confidence signals, source attribution, and explainable outputs.

Human-in-the-Loop+

Designed the handoff moments between AI and human — review flows, correction patterns, escalation logic, and audit trails.

Enterprise Product+

End-to-end product design for complex B2B platforms — from discovery and research through to production-ready specs.

Conversational UI+

Design chat interfaces, copilot experiences, and guided AI flows that feel natural — balancing automation with user control.

Strategy+

Connect design decisions to business outcomes — translating ambiguous stakeholder requirements into clear, testable solutions.

01 / 06
Agentic UX Design
AI Transparency
Human-in-the-Loop
Enterprise Product
Conversational UI
Human TrustAI + human judgmentResearchneeds_mappedevidence_linkedPrototypeprototype_readyflow_testedStrategyproblem_framedtradeoffs_clearTrustreview_visiblehuman_control

Selected UX & AI Projects.

Industry Experience.

Warner Bros. DiscoveryMedia & entertainment
Enterprise AI

Multi-agent systems for Warner Bros. Discovery, as described in my portfolio.

Portfolio context
SanofiPharma & regulatory
Regulatory AI

Regulatory AI work for Sanofi. Each project taught me that trust is a design deliverable — not a feature.

Portfolio context
Rice GroupEducation
Industry experience

Education is one of the industries I have worked in. The core challenge is making AI trustworthy for non-technical users.

Portfolio context
Legal & compliance
Enterprise knowledge platforms

Enterprise knowledge platforms used by legal and compliance teams globally. I focus on confidence signals, error states, and correction flows.

sumitai.work
Human-centred AI
Portfolio context

Senior Product Designer specialising in agentic systems, enterprise AI, and human-in-the-loop experiences.

Portfolio context

Source Outcomes

RegIntel AI · source-reported

SUBMISSION PREP TIME · SOURCE REPORTS
−55%
Before launch100
After launch45

Relative prep time · baseline = 100.

RegIntel AI · audit workflow

AUDIT PREP TIME · SOURCE REPORTS
−65%
Before launch100
After launch35

Relative audit prep · baseline = 100.

Frequently Asked Questions

What kind of roles are you looking for?

Senior Product Designer roles focused on AI products, agentic systems, and enterprise UX — primarily in Dubai. Open to full-time positions with teams building complex, human-centred AI experiences.

What industries have you worked in?

Media and entertainment (Warner Bros. Discovery), pharmaceutical and regulatory (Sanofi), education ( Rice Group), and enterprise knowledge management. Across all of them, the core challenge was the same — making AI trustworthy for non-technical users.

What does your design process look like?

I start with the problem, not the screen. Research, stakeholder alignment, and edge case mapping before any wireframe. For AI products specifically, I map the full system — agent behaviour, confidence states, human handoff moments — before designing the interface on top of it.

Are you open to relocation?

Yes — open to opportunities in Dubai and across India. Available to join immediately.

Have you worked directly with AI engineers and data scientists?

Yes — on every AI project. I work embedded with engineering teams, not handed off to them. That means I understand model confidence, pipeline constraints, and what's technically feasible — which makes my designs ship at high fidelity without constant revision cycles.

( LET'S TALK )

Let's Design Something
Incredible Together.

Book a Call