Junior Full-Stack Engineer - LLM-Powered Products x2

Established in Canada in 2015, Valsoft has grown to a global portfolio of 118+ companies, acquiring and developing vertical market software companies, enabling each business to deliver the best mission-critical solutions for customers in their respective industries. A key tenet of Valsoft’s philosophy is to invest in well-established businesses and foster an entrepreneurial environment that molds companies into leaders in their respective industries. Valsoft looks to buy, hold and create value through long-term partnerships with existing management. 

Requirements

What You’ll Do

Build user-facing features in React/Next.js and hook them to back-end APIs you write in TypeScript (Node) or Python (FastAPI). • Craft and version prompts; experiment with tools like LangChain to chain calls and log outputs for review.

• Use AI pair-programming tools (Cursor, Windsurf, Copilot, etc.) to speed up coding, refactors, and test generation.

• Spin up deployments via Vercel / Fly.io or GitHub Actions, debug when things inevitably break, and celebrate the green check-mark.

• Join product brainstorms—your side-project instincts are welcome.

• Keep an eye on basic metrics (load time, error rate) and squash bugs before users tweet them.

Must-Haves

• Solid grasp of JavaScript/TypeScript and React plus comfort in either Node or Python back-end work.

• Comfort with AI code editors—show us a commit, video, or repo where you actually used one.

• Basic Git workflows: branch → PR → merge, and you know why CI turns the build red. • Familiarity with any deploy flow (Vercel button, Fly.io, Heroku, GitHub Actions, etc.).

• Curiosity for how products delight users, not just how code compiles.

• Proven fluency in working with REST/GraphQL APIs—comfortable crafting and parsing JSON payloads, debugging requests with Postman/Insomnia or browser dev-tools, and handling error codes/retries.

• Hands-on experience iterating prompts in tools like ChatGPT, Cursor, or Windsurf. You can show how prompt tweaks changed an LLM’s output to match a specific product need. Nice-to-Haves

• LangChain, LlamaIndex, or any RAG experiment on your GitHub.

• Vector database dabbling (pgvector, Pinecone).

 • A side project people outside your family have used. Our Stack React 18

• Next.js

• Node 20

• FastAPI

• OpenAI & Anthropic APIs • Postgres + pgvector

• Vercel

• Fly.io

• GitHub Actions

• Cursor & Windsurf for daily coding Hiring Process (max 10 days) 1. 30 min intro call – culture & salary chat, bring questions. 2. 2 hr take-home – build a tiny LLM-powered feature and push via GitHub Actions (we pay £50 for your time). 3. 60 min tech talk – walk through your code, refactor a piece live (AI editor allowed). 4. Founder chat – product vision and growth path. 5. Offer – within 48 hrs.

This role is Hybrid London (2 days/wk in London) with a M-series MacBook, £1k annual learning budget and a 5% bonus of salary

Why You’ll Love It

• Ship real stuff fast – small team, weekly releases, instant user feedback.

• Play with cutting-edge toys – OpenAI, Anthropic, or custom LLM endpoints, vector stores, and AI code editors (Cursor, Windsurf, Copilot).

• Mentorship that matters – direct pairing with a senior lead; you’ll see how product decisions are made, not just tickets.

• Clear growth path – successful product launch will lead to a clear path to mid-level within ~18 months. What You’ll Do

• Build user-facing features in React/Next.js and hook them to back-end APIs you write in TypeScript (Node) or Python (FastAPI).

• Craft and version prompts; experiment with tools like LangChain to chain calls and log outputs for review.

• Use AI pair-programming tools (Cursor, Windsurf, Copilot, etc.) to speed up coding, refactors, and test generation.

• Spin up deployments via Vercel / Fly.io or GitHub Actions, debug when things inevitably break, and celebrate the green check-mark.

• Join product brainstorms—your side-project instincts are welcome.

• Keep an eye on basic metrics (load time, error rate) and squash bugs before users tweet them.

Closing date for this role is 18th August 2025