Itzik Bitran

Before Franklin, there was Genoox: a variant-analysis platform sold only to laboratories, unusable without a purchase. To reach the individuals asking about their own results, we extracted just the classification engine and put it online, free. It found its market immediately. The gradient bar we designed for it became recognizable enough that competitors later copied it, brand and mascot included.

Chapter 1 — Trusting an algorithm with a diagnosis

A classification is only useful if the person reading it trusts it enough to act. Franklin runs a variant through the ACMG rule set automatically, but a reviewer can open any rule, see the evidence behind it, adjust it, or override it with something only they would know, a specific ethnicity, a case they've seen before. That combination was the plan from day one, not a fix added later: the algorithm doesn't replace the reviewer, it gives them a documented place to start, and disagreement is surfaced rather than hidden.

Chapter 2 — Onboarding with no onboarding

The free tool needed to work for lab directors and physicians, and equally for someone who just received a genetic report and wants to know what their own variant means. Nobody would be trained on it. The fix was two decisions: show results in fractions, an aggregated view first with every layer beneath it optional to open, and build the flow entirely from patterns people already know, a search bar, a folder, an email list, rather than inventing new ones.

The free calculator did more than get used: it became the reason people asked their own labs to buy the full platform, a direct, named mechanism rather than a vague growth claim.

ThetaRay came to Uniq UI with a working model and nothing else. I designed the product end to end, UX and UI, starting hands-on and later directing the engagement, and built its visual brand alongside it, since none existed.

The problem underneath every screen: an accurate anomaly score means nothing if the analyst reviewing it doesn't trust it enough to act. The product team thought in mathematical terms, not user needs, and there was no library of "solved" screens anywhere in the field to draw from, every other platform looked like an extended spreadsheet.

Configuration, without a data scientist

Detection and drill-down

Presented at a major industry conference to excellent feedback that helped ThetaRay secure funding, alongside the brand identity built in parallel with the product.

Sixgill had nothing before Uniq UI: no product, no brand. I led the design concept and brand, started the build, then handed execution to my team, the same hands-on-then-delegate pattern as ThetaRay. The end user is a threat-intelligence researcher scanning the dark web for early signals, hints and trends in a huge, unstructured firehose of posts.

The hard part was volume: presenting all of it without overwhelming the reader, solved with an aggregated-summary-first approach. The breakthrough was the Social Network chart above, a force-directed graph of an actor's connections that became the platform's most recognizable, best-loved element, for users and for Sixgill's own leadership. The dark, neon-accented brand was a deliberate match for the subject matter.

It's worth setting this next to ThetaRay: two different visualizations solving the same underlying problem, no existing visual precedent for a new kind of data, a cluster map for anomaly scores, a network graph for actor relationships.

UI across the whole ecosystem was mine; UX was led by someone else, worth saying plainly rather than claiming both. Two audiences: guests booking a visit and buying gifts in advance, and DreamWorks' own operations team managing bookings and gift selections day to day. The only piece that existed before was the gift-buying flow, built on DreamWorks' e-commerce partner Jifiti; everything else, marketing site, scheduling, the operator platform, was new.

The genuinely hard part: four apps needed to share data under a schedule that didn't move, and keeping one consistent brand across a high volume of screens produced fast meant constant revision. The slowest link was distance itself, DreamWorks in the US, Uniq UI in Israel, feedback cycles stretched thin against a big brand's high bar for quality.

Every screen was also built fully responsive, desktop, tablet, and mobile alike, since guests booked visits and browsed the gift catalog from whichever device they had in hand, not just one. The three shots here are the same ecosystem across all three.

Shipped on time with minimal customer-facing bugs, and ran successfully for a couple more seasons. Backed by public references, not just an account of it: