AI voice agent platform
CallGenie
A dashboard for running AI voice agents: upload contacts, schedule outbound call campaigns, capture bookings and review call performance.
Next.jsReactTypeScriptPostgreSQL
Read case study
AI features & integration
AI developer for web applications: LLMs, voice agents and vision
I'm a full-stack and AI developer based in Sri Lanka, working remotely with businesses abroad. AI is useful when it is built into a workflow people already have, with sensible limits and an interface that explains what it is doing. I have built LLM-powered writing tools, AI voice agent platforms and image classification features into full applications — and I focus on the product around the model as much as the model itself.
Who this is for
- SaaS products adding AI-assisted features for their users
- Businesses automating repetitive calls, writing or classification tasks
- Founders whose product idea depends on an AI capability
Problems this solves
- You want to add AI but are unsure which use case is worth building first
- A proof-of-concept works in a notebook but not inside your product
- AI costs are unpredictable without usage limits
- Users do not trust AI output they cannot understand
What is included
- Use-case review to find where AI genuinely helps
- Integration with LLM APIs such as Google Gemini
- AI voice agents through platforms such as Vapi
- Image classification model serving through a Python service
- Usage controls: credits, rate limits and input validation
- Interfaces that explain AI output and its limits
Technical capabilities
- LLM APIs and prompt design
- Voice AI platforms (Vapi)
- Python inference services
- Vision models (DINOv2, Vision Transformer)
- Next.js and TypeScript product integration
What you receive
- AI feature integrated into your application
- Usage and cost controls
- Documentation of prompts, models and configuration
What affects scope and cost
Every quote is based on a written scope. These are the factors that usually move it the most:
- Whether an off-the-shelf model API is enough or a custom model is required
- Data availability and quality
- Expected request volume and cost limits
- Privacy and data-handling requirements
How to get started
- 01Send a short description of your project through the inquiry form — a few sentences is enough.
- 02We have a 30-minute video call at a time that suits your time zone.
- 03You receive a written scope with milestones and a quote for each.
Related work
AI writing SaaS
ResearcherAI
A credit-based SaaS that helps students and researchers draft paper sections, improve drafts, paraphrase and generate citations in IEEE, Springer and ACM formats.
Next.js 14TypeScriptGoogle GeminiPostgreSQL
Read case study
Pet healthcare platform
VetLink
A pet healthcare web app with image-based dog skin condition detection, limping detection, health assessment and plain-language AI guidance for owners.
Next.jsTypeScriptPythonDINOv2
Read case study
Questions about ai features & integration
Do I need my own AI model?
Usually not. Most product features work well on existing model APIs. Custom training only makes sense when you have specific data and a task general models handle poorly.
How do you keep AI costs under control?
With input validation, rate limits and usage-based limits such as credits, so every request has a known ceiling.
What happens to our data?
We agree up front which data is sent to which provider. Sensitive data can be minimised or excluded, and provider settings are chosen to match your requirements.
Have a project in mind?
Tell me what you are building and where you are stuck. I will come back with questions or next steps — and if I am not the right fit, I will say so.