19+ years shipping AI-powered SaaS, enterprise systems, e-commerce platforms, automation workflows, and high-performance web applications with TypeScript, Next.js, React, Node.js, Python, PostgreSQL, Prisma, and OpenAI APIs.
Every figure below traces back to a shipped system, a profiling run, or a client report — never to a slide deck.
Senior engineering across SaaS, enterprise systems, e-commerce, AI workflows, and infrastructure.
01Launched web properties across insurance, commerce, automation, and content-heavy products.
02Built production web apps, internal tools, automation systems, and AI product prototypes.
03Insurance-sector digital sales impact from shipped e-commerce and workflow systems.
04Performance improvements through profiling, caching, frontend tuning, and infrastructure work.
05Broad training across software engineering, cloud, DevOps, security, APIs, testing, and agile delivery.
06The site mirrors the CV: broad senior delivery, concrete performance outcomes, AI product depth, and remote collaboration across SaaS, commerce, insurance, and enterprise workflows.
Prishtina, Kosovo / Remote — with UK, Sweden, insurance, e-commerce, and SaaS delivery history.
MSc and BSc in Management Information Systems; Albanian native and English C2 professional proficiency.
Load-time budgets, cost-per-request, and error budgets are agreed at the start, not negotiated after launch.
Architecture, implementation, DevOps, and the user-facing screen owned by the same senior pair of hands.
“He took over a stalled build, shipped the missing 30%, and left us docs we actually use.”
Engineering lead · UK e-commerce group · 2023
Architecture, implementation, performance, DevOps, automation, and user-facing systems — kept connected instead of handed across a wall.
Before architecture, the workflow. What enters the system, who touches it, where it stalls, and what evidence proves it worked. Constraints become requirements.
Data model first, then contracts, then UI. Nothing speculative gets built. The first release has to survive real operators on a bad Tuesday.
Weekly demo-able increments on preview deployments. Every slice is testable by the people who will actually use it, with feedback folded into the next one.
Load times, query plans, token cost, error budgets. Profiling before optimisation, and numbers recorded before and after so the claim can be checked.
Runbooks, architecture notes, recorded walkthroughs, and a codebase another team can read. Dependency on me is a failure mode, not a business model.
This is the delivery shape behind the CV outcomes: map the operating problem, build the system, verify performance, and leave useful handoff material.
Map the operating problem, users, and the bottleneck worth solving.
Schema, contracts, UI, integrations — typed end to end, shipped in slices.
Profile, tune, and record before/after numbers as evidence.
Runbooks, architecture notes, walkthroughs, and a readable codebase.
End-to-end AI products: retrieval, agents, structured outputs, evaluation, billing, and the boring reliability work that keeps them alive after launch week.
TypeScript across Next.js, React, and Node with PostgreSQL and Prisma. Design systems, API contracts, auth, and deployment handled by the same person who writes the schema.
Workflow software around the way your team actually operates: intake, routing, verification, notifications, and the audit trail that proves it ran.
Profiling slow apps, databases, and build pipelines. Real numbers before real changes — caching, query plans, bundle budgets, and infrastructure tuning.
Architecture reviews, code health passes, deployment hardening, and documentation so an internal team can carry the system forward without a dependency on me.
Scoping an AI feature is arithmetic, not vibes. The Bench holds working versions of the calculations behind every estimate on this site — priced from published rates, scored against real thresholds. They run entirely in your browser.
“Shipped the missing 30% in six weeks and left docs we still use.”
“Tender analysis went from three days of reading to an afternoon of deciding.”
“Argues about query plans in the morning and button hierarchy in the afternoon.”
“Shipped the missing 30% in six weeks and left docs we still use.”
“Tender analysis went from three days of reading to an afternoon of deciding.”
“Argues about query plans in the morning and button hierarchy in the afternoon.”
“The only contractor who sent us a performance report we could verify.”
“Our reception load dropped and the call records finally became usable data.”
“Handover was so clean we forgot he was external within a month.”
“The only contractor who sent us a performance report we could verify.”
“Our reception load dropped and the call records finally became usable data.”
“Handover was so clean we forgot he was external within a month.”
Explore practical guides on AI tools, structured outputs, performance, and reliable workflows.
Design the identity, retrieval, citation, and test boundaries for a document assistant used by more than one customer.
Turn retrieval results and file annotations into a source experience that readers can inspect without overstating certainty.
Create a small, versioned set of agent tests that checks behavior, evidence, permissions, and recovery—not just the final wording.
Send the desired outcome, the users, the current bottleneck, the stack, and the timeline. You'll get a practical senior engineering path for the build — not a brochure.
