AI integration for web apps (LLMs, chat, speech-to-text)Amadou Boubacar Niang
I add AI to your product in a way that holds up in production: writing assistants, summaries, chat over your content, AI-drafted support replies and speech-to-text transcription.
Each feature uses the cheapest model that does the job, falls back to another provider when one fails, and runs under a daily spending cap, so an outage or a bug can't run up a surprise bill.
What I can build
AI writing assistants and content generation
Summaries, chat and search over your documents or data
AI-drafted support replies, with a person approving before anything is sent
Transcription with Whisper (self-hosted on a GPU or through an API), with speaker labels and exports
Integrations with Gemini, DeepSeek, OpenAI-compatible APIs or local models
What you get
A recommendation on models and costs before we build
The feature built into your app, with fallbacks and spending caps
Tests and a short write-up of how it works
Timeline
1 to 3 weeks for a single feature. 4 to 8 weeks for a full AI product, such as a transcription app.
What I need from you
The feature you have in mind and access to your codebase, or a brief if we're starting from scratch.
Proof
Built in production: the AI press release writer and support desk on PRNow, and Podcast Transcript AI, which runs on self-hosted Whisper with 52,700+ public transcripts in 73 languages.
Contact for pricing
Duration1 week
Tags
Google Gemini
Next.js
Node.js
AI Engineer
Backend Engineer
Fullstack Engineer
Service provided by
Amadou Boubacar Niang St Louis, Senegal
AI integration for web apps (LLMs, chat, speech-to-text)Amadou Boubacar Niang
Contact for pricing
Duration1 week
Tags
Google Gemini
Next.js
Node.js
AI Engineer
Backend Engineer
Fullstack Engineer
I add AI to your product in a way that holds up in production: writing assistants, summaries, chat over your content, AI-drafted support replies and speech-to-text transcription.
Each feature uses the cheapest model that does the job, falls back to another provider when one fails, and runs under a daily spending cap, so an outage or a bug can't run up a surprise bill.
What I can build
AI writing assistants and content generation
Summaries, chat and search over your documents or data
AI-drafted support replies, with a person approving before anything is sent
Transcription with Whisper (self-hosted on a GPU or through an API), with speaker labels and exports
Integrations with Gemini, DeepSeek, OpenAI-compatible APIs or local models
What you get
A recommendation on models and costs before we build
The feature built into your app, with fallbacks and spending caps
Tests and a short write-up of how it works
Timeline
1 to 3 weeks for a single feature. 4 to 8 weeks for a full AI product, such as a transcription app.
What I need from you
The feature you have in mind and access to your codebase, or a brief if we're starting from scratch.
Proof
Built in production: the AI press release writer and support desk on PRNow, and Podcast Transcript AI, which runs on self-hosted Whisper with 52,700+ public transcripts in 73 languages.
Contact for pricing