{"$schema":"https://jsonresume.org/schema/1.0.0/resume.json","basics":{"name":"Pavlo Puzikov","label":"Innovation researcher · applied AI · product","image":"https://pavlopuzikov.com/visuals/pavlo-portrait.jpg","email":"pavlopuzikov@gmail.com","summary":"An innovation and product researcher working across immersive 3D, applied AI, and generative video. The work spans a real-time Gaussian-splat research pipeline, directing generative-video launch films, earning brand visibility inside AI answer engines, and representing brands on-stand at global tech expos. The throughline is taking an emerging technique from first experiment to something a business can put in front of a client.","location":{"city":"Dubai","countryCode":"AE","region":"Dubai"},"profiles":[{"network":"LinkedIn","username":"pavlo-puzikov","url":"https://www.linkedin.com/in/pavlo-puzikov"},{"network":"GitHub","username":"pavlopuzikov","url":"https://github.com/pavlopuzikov"},{"network":"Letterboxd","username":"pavlemio","url":"https://letterboxd.com/pavlemio/"}],"url":"https://pavlopuzikov.com"},"work":[{"name":"BARNES Dubai","position":"Product Innovation Researcher","startDate":"2024","summary":"Product Innovation Researcher at BARNES Dubai, concurrent role.","url":"https://pavlopuzikov.com"},{"name":"Living Homes","position":"Marketing & Partnerships","startDate":"2024","summary":"Marketing & Partnerships at Living Homes, concurrent role.","url":"https://pavlopuzikov.com"}],"projects":[{"name":"Barnes Vantage","description":"One map-led surface that folds the scan library, the live 3D viewer, and the listing intelligence into a single tool instead of four. Every point on the map is a real on-site capture; pick one and walk it in the browser.","highlights":["Folds the standalone Splat Viewer, the scan library, and the valuation signals into one dark, map-led surface a broker opens instead of four tools. In development: the map, the natural-language search, and the rankings, predictions, and experiments panels run against live scan data, with the same walkable scan engine embedded at the centre.","primary: Map-led Next.js interface: a MapLibre canvas, natural-language search over the listings, and the rankings, predictions, and experiments panels driven by live scan data.","secondary: The walkable scan at the centre, the same gaussian-splats-3d viewer engine embedded per residence.","secondary: Folding four separate broker tools (library, viewer, valuation, experiments) into one operator surface.","supporting: Scan-coverage and value signals computed over the listing data behind the panels."],"keywords":["Next.js","MapLibre GL","Three.js","gaussian-splats-3d","Cesium","Typesense","Prisma"],"url":"https://pavlopuzikov.com/work/splat-viewer","roles":["Owner: product, design, and full-stack build."],"entity":"Studio","type":"3D & Spatial","status":"In development"},{"name":"Generative Video & Creative","description":"Directing generative-video films end to end, from brief and storyboard to directing the shots to a delivered edit, working closely with talented videographers and editors, alongside creative direction across BARNES Dubai and Living Homes. The Marina Sozopol launch film ran to 27 storyboarded shots blending real footage, 3D renders, and AI-generated video; the In motion series carries the brand films, with the Maison Margiela Dubai Residences launch as the flagship.","highlights":["primary: Directing generative-video films end to end: brief, storyboard, shot list, directing AI-generated shots, curation, and edit, across client-feedback rounds.","secondary: Marketing direction for luxury branded-residences launches: positioning, content strategy, and campaign ideation with the marketing team.","secondary: Producing shots with image-to-video generation off generated key frames, matching a fast model to simple shots and a stronger one to complex.","supporting: Content ideation, review, and on-brand QA to a Maison Margiela-grade luxury standard."],"keywords":["Storyboard & shot list","Image-to-video","Edit decision list","Creative direction","Content strategy"],"url":"https://pavlopuzikov.com/motion","roles":["Generative-video direction, creative direction, content strategy, and production with the marketing team."],"entity":"Studio","type":"Marketing","status":"In production"},{"name":"3DGS Research Pipeline","description":"20 Gaussian-splat engines integrated into one benchmark harness, 6 evaluations run so far. Which one actually survives a real luxury-interior scan, not just the leaderboard.","highlights":["A reproducible benchmark harness: around twenty engines wired, 6 evaluations run so far, a scan-difficulty taxonomy and per-engine failure modes, that the academic write-up draws on. The full engines-by-scans matrix and the papers are in progress. A relighting track now runs alongside the benchmark: an interactive viewer with movable light fixtures and real soft cast shadows, plus a delighting and relightable-retrain line still in progress.","primary: Around twenty 3DGS engines wired into one harness spanning studio scans to city-scale capture, with the first evaluations run on PSNR, SSIM and LPIPS and the full matrix in progress.","secondary: Novel-view synthesis and monocular depth paths for unbuilt properties stitched into the same pipeline.","secondary: Video-diffusion novel-view models evaluated against COLMAP-grounded baselines.","supporting: Reproducible eval harness across PSNR, SSIM, and per-engine failure modes."],"keywords":["CUDA","gsplat","PyTorch","COLMAP","Video diffusion","Monocular depth"],"url":"https://pavlopuzikov.com/work/3dgs-research","roles":["Researcher / implementer: around twenty engines wired across the pipeline."],"entity":"Studio","type":"3D & Spatial","status":"In development"},{"name":"Data Atlas","description":"An entity-centric valuation platform and on-demand AVM over 538K entities across 28 sources, with a MAPE and RMSE calibration and backtest harness; an internal snapshot showed ~7% MAPE.","highlights":["Turns multi-day analyst cycles into an on-demand AVM brief, ranked opportunities, and an accuracy snapshot the valuation team can audit.","primary: FastAPI service on PostgreSQL with hex-grid spatial index, confidence-tiered entity ontology, and entity resolution across 538K entities from 28 sources.","secondary: Automated valuation model with a MAPE, RMSE and calibration backtest harness (an internal snapshot showed roughly 7% MAPE), six-month community-calibrated forecasts, and a six-dimension risk-adjusted investment score.","secondary: 28 sources reconciled into one analyst surface, from DLD transactions / buildings / brokers / projects / developers / permits / rent / freehold / land to Barnes Internal LLM market, Dubai-coordinate geo, and more. Real Dubai Land Department transaction data, zero mock.","secondary: React + Cesium AVM tool + DLD-comparables panel + accuracy snapshot dashboard. The customer-facing valuation surface the team opens daily."],"keywords":["Python","FastAPI","PostgreSQL","Cesium","React","Machine-learning pipelines"],"url":"https://pavlopuzikov.com/work/atlas","roles":["Owner, architecture, data, machine learning, AVM tool, frontend."],"entity":"Studio","type":"AI / ML","status":"In production"},{"name":"Barnes Dubai LLM","description":"A domain-tuned LLM and agentic broker assistant for Dubai luxury real estate, a QLoRA fine-tune on 15,369 instruction pairs, running locally.","highlights":["Live in the Innovation Portal playground. A broker asks `Compare Business Bay vs JVC for buy-to-let yield` and gets a tool-resolved verdict in seconds (medians, gross yields, tenant pool, a one-paragraph recommendation), the brief that used to take an analyst an afternoon.","primary: QLoRA domain fine-tune + a tool-calling agent loop (getCommunityInventory, comp-search, valuation) with audit-logged human-in-the-loop above the confidence threshold.","secondary: Prompt + tool-schema design for the agent loop, system prompts, function-calling contracts, a confidence-gated escalation prompt, and an eval harness over real broker queries.","secondary: FastAPI inference service with broker_id-keyed conversation memory; Ollama deployment; WhatsApp surface in front of the playground.","secondary: 15,369 analyst-reviewed instruction pairs across five national registries, plus live Atlas connectors feeding tool calls.","supporting: Training-run analysis and dataset assembly that targeted brokerage-grade jurisprudence."],"keywords":["PyTorch","QLoRA","Ollama","Python","FastAPI","Tool calling","Multi-agent"],"url":"https://pavlopuzikov.com/work/barnes-ai","roles":["Owner, dataset, training, tool-calling agent loop, inference stack, broker playground."],"entity":"Studio","type":"AI / ML","status":"In production"},{"name":"Academic Papers","description":"Two companion academic papers on applied Gaussian splatting for luxury interiors, one on production-deployment lessons from the benchmark, one (Reflection-Aware 5DGS) on the mirror-and-glass problem. Both drafted; a Tech Communications submission in preparation.","highlights":["primary: Sole author of two companion academic papers: production-deployment lessons from the benchmark, and Reflection-Aware 5DGS (classified-quadric reflectors for specular interiors).","secondary: Engine benchmark and scan-difficulty taxonomy from Easy Studio through Extreme Commercial Office.","secondary: Novel-view synthesis comparison across diffusion-based and photogrammetric baselines.","supporting: PSNR / SSIM heatmaps and per-engine robustness scoring."],"keywords":["3DGS","Novel-view synthesis","Video diffusion","Research writing"],"url":"https://pavlopuzikov.com/work/deploying-3dgs","roles":["Sole author, two companion academic papers, in preparation."],"entity":"Studio","type":"Research","status":"In development"},{"name":"Innovation Portal","description":"Internal cockpit for the BARNES Dubai innovation team: project tracking, multilingual RBAC (EN / FR / AR), embedded Apache Superset BI, and a marketing-engineering toolkit (SEO/AEO/GEO, competitor console, Data Atlas, 3D Scans, CRM Copilot, an in-portal Barnes Dubai LLM assistant, and an AI content studio for marketing scripts and storyboards).","highlights":["Live internally, the single surface from which the BARNES Dubai team runs 23 innovation projects, 170 milestones, 345 tests, and the AEO/SEO + Barnes Dubai LLM tooling.","primary: Designed the cockpit IA, OVERVIEW (Dashboard / Projects / Updates), KNOWLEDGE (Data Atlas / 3D Scans / Documents), TOOLS (Barnes Dubai LLM, AEO/SEO, Translate, CRM Copilot, Uptime).","secondary: Next.js + React app with dense data UI, multilingual RBAC, and embedded Superset analytics surfaces.","secondary: Auth, role gating, project + milestone + test telemetry pipelines, and integration with the Barnes Dubai LLM inference stack.","supporting: AEO/SEO/GEO tool surface: keyword + prompt research, generative-engine visibility scoring, AI-answer-engine tracking across ChatGPT / Claude / Perplexity / Gemini.","supporting: Creative content QA and ideation for the marketing toolkit, reviewing AI-drafted copy and AEO answers before they ship."],"keywords":["Next.js","React","TypeScript","Apache Superset","PostgreSQL","RBAC","i18n"],"url":"https://pavlopuzikov.com/work/innovation-portal","roles":["Owner, product, architecture, build, deployment."],"entity":"Studio","type":"Product","status":"In production"},{"name":"House Call","description":"Invitation-only natal-reading practice. Hand-prepared readings synthesise Western, Chinese BaZi, and Vedic Jyotish, backed by per-user accounts, daily astronomy-engine transits, a falsifiable prediction tracker, synastry, and tarot import.","highlights":["Live at housecallastro.com as an invitation-only private beta, payments off. The public launch and pricing follow once the founders' cohort closes.","primary: Synthesised three-tradition reading practice with falsifiable prediction tracking; invite-only by design.","secondary: Next.js app with per-user auth, daily transit forecasts, synastry, and tarot import surfaces.","secondary: Turso libSQL on Prisma, jose-signed JWTs, astronomy-engine transit cron, Vercel-fronted deployment.","secondary: Midnight + gold editorial identity with a gold orbital glyph that anchors the landing."],"keywords":["Next.js","Prisma","Turso","JWT","astronomy-engine","Vercel"],"url":"https://pavlopuzikov.com/work/threadwork","roles":["Owner, product, practice, content, deployment."],"entity":"Personal","type":"Product","status":"In beta"},{"name":"AriOS","description":"A personal knowledge companion that captures what you see, read, and hear into an Obsidian vault, then helps you actually learn it. A Knowledge Trainer turns notes into flashcards, grades your free recall, and schedules spaced-repetition reviews. Local-first; vault sovereignty is the hard constraint.","highlights":["Daily driver, now a learning engine too. Capture and the trainer's auto-cards + graded recall are shipped; server-side spaced repetition with Telegram reminders is the active build.","primary: Pluggable multimodal pipeline: vision (Cloudflare llava / Ollama with stub fallback), transcription (Cloudflare Whisper), classification (Ollama with heuristic fallback), retrieval (BM25 + MiniLM + CLIP via RRF), vault RAG with citation.","supporting: Prompt design for the on-device classifier and the vault-RAG /ask path, heuristic-first routing with an LLM fallback prompt tuned for low latency and citation.","secondary: One Express server, five capture surfaces. URL ingest with platform-specific extractors (yt-dlp, Mozilla Readability, Reddit JSON API). Async job queue with status polling. Vercel-hosted Telegram webhook bot. Vercel-cron reminders running 24/7.","secondary: Vault sovereignty as a hard constraint (markdown only, no DB, no sidecar). Telegram as the universal capture surface, replaces iOS Shortcut maintenance. Heuristic-first classify so the loop never blocks on a slow LLM. Pluggable provider layer so cloud paths are opt-in, not load-bearing.","supporting: Append-only JSONL retrieval index next to the vault. Activity log per ingest with per-step timing. Full source preserved in collapsed <details> blocks so the index has the raw words, not just the summary."],"keywords":["TypeScript","Express","Telegram Bot API","Cloudflare Workers AI","Ollama","Whisper"],"url":"https://pavlopuzikov.com/work/ari","roles":["Owner, architecture, every capture surface, pluggable vision + classify layers, retrieval, daemons."],"entity":"Personal","type":"AI / ML","status":"In beta"},{"name":"Splatlas","description":"A navigable solar system where every world is a real 3D Gaussian splat, trained on a GPU from open NASA and USGS imagery and elevation, not a texture on a sphere. Fly through, warp to any planet, read it up close.","highlights":["Live at splatlas.space: the Sun to Neptune as navigable splats with real orbits, a date scrubber, and live space-weather, all running in the browser.","primary: Every body is a real 3D Gaussian splat trained from satellite imagery and elevation, not a textured sphere.","secondary: Three.js flythrough: warp-to-planet navigation, a true-scale toggle, orbital motion, and a date scrubber, all real-time in the browser.","secondary: Pipeline turning open NASA and USGS imagery and DEMs into per-body point clouds and trained splats, Sun to Neptune.","supporting: Applied 3DGS past capture: emission-baked splats and per-body lighting for worlds you cannot photograph in the round."],"keywords":["3D Gaussian Splatting","Three.js","gsplat","WebGL","NASA / USGS DEM","Python"],"url":"https://pavlopuzikov.com/work/splatlas","roles":["Solo: the data pipeline, the splat training, and the Three.js flythrough app."],"entity":"Personal","type":"3D & Spatial","status":"Live"},{"name":"Smart Deal Global","description":"A public marketplace for luxury real-estate opportunities, multilingual (EN / AR / FR / RU), with hreflang and structured data tuned for search and answer-engine discovery. I built the first version of the site; a talented colleague has since drastically upgraded it.","highlights":["Lead capture via WhatsApp, phone, and email to sales advisors.","primary: Hreflang plus structured-data optimisation for search and answer-engine discovery across EN / AR / FR / RU.","secondary: Creative content QA and ideation across four languages, headline and listing-copy review, on-brand tone, and campaign concepts for the marketplace.","secondary: Editorial landing with Dubai-skyline hero, GSAP scroll motion, and Swiper-driven listing galleries.","secondary: Next.js + i18n stack with EmailJS / WhatsApp / phone lead-capture wired to sales advisors.","supporting: Curated copy and visual language matched across four languages and ten markets."],"keywords":["Next.js","i18n","Structured data","EmailJS","GSAP","Swiper"],"url":"https://pavlopuzikov.com/work/smart-deal-global","roles":["Initial design, copy, and build (first version)."],"entity":"Studio","type":"Marketing","status":"Live"},{"name":"SEO · AEO · GEO Toolkit","description":"An audit, scoring, and first-party measurement toolkit for staying visible in AI answers (ChatGPT, Perplexity, Claude) and in Google, scored from your own Search Console data, not a third-party broker's.","highlights":["Live in the BARNES Innovation Portal: paste any URL and get a real on-page audit, separate SEO and AEO/GEO readiness scores, a simulated answer-engine visibility check, and an audit-grounded action plan. A companion first-party layer reads Google Search Console and GA4 locally and returns quick wins (queries ranking 5 to 15), topic clusters, content gaps, title and meta CTR fixes, and a weekly report, so one instrument covers both what AI answers cite and what real search traffic does.","primary: A research playbook separating SEO (Google), AEO (answer-engine citation), and GEO (generative-engine) visibility, turned into an audit + scoring toolkit a marketing team can run on any URL.","secondary: Distilled how answer engines decide what to cite, and how differently that is from how Google ranks, into the handful of signals worth acting on.","secondary: A web audit surface that reads a live page the way a crawler does, returns separate SEO and AEO scores, and keeps a history the marketing team works from.","supporting: A visibility simulation that estimates how a brand surfaces in conversational answers, plus an audit-grounded action plan, built on swappable, mostly-local model providers.","secondary: A first-party measurement layer that reads Google Search Console and GA4 locally, via Google's own APIs with a CSV-paste fallback, and turns the data into quick wins, query clusters, content gaps, title and meta CTR fixes, and a weekly report."],"keywords":["Next.js","TypeScript","Python","pandas","Prisma","Ollama","Search Console API","GA4","Brave Search API"],"url":"https://pavlopuzikov.com/work/ai-search-visibility","roles":["Owner, research playbook, toolkit, scoring model, dashboard."],"entity":"Studio","type":"Marketing","status":"In production"}],"skills":[{"name":"Applied AI","level":"Expert","keywords":["QLoRA fine-tune","VLM (Ollama, Anthropic)","Multimodal retrieval (CLIP + MiniLM + BM25 + RRF)","Multi-agent orchestration"]},{"name":"3D and spatial","level":"Expert","keywords":["3D Gaussian Splatting (gsplat, gaussian-splats-3d)","COLMAP / photogrammetry","Three.js / WebGL","Cesium"]},{"name":"Frontend engineering","level":"Proficient","keywords":["Next.js","React","TypeScript","Tailwind","motion/react"]},{"name":"Backend engineering","level":"Proficient","keywords":["FastAPI","Express","PostgreSQL","Prisma","Turso (libSQL)"]},{"name":"Marketing / AEO / GEO","level":"Proficient","keywords":["Answer-engine optimisation","Structured data (Schema.org)","Hreflang and multilingual","AI-answer-engine tracking"]}],"languages":[{"language":"English","fluency":"Native or bilingual"},{"language":"Russian","fluency":"Native or bilingual"}],"meta":{"canonical":"https://pavlopuzikov.com/resume.json","version":"1.0.0","generated":"2026-07-11T09:51:37.259Z","source":"Generated from content/data/work.ts and content/data/profile.ts in pavlopuzikov/pavlo-portfolio. The HTML CV lives at https://pavlopuzikov.com/cv."}}