AI Model Integration Jobs

The goal of this task is to train and finetune the Gemma4 Model. I already have a sample training dataset consisting of -incoming mesage -thinking steps -desired final reply I want a trainings pipeline to finetune the Gemma4 model for this datasetp. It should optimize both, thinking and and final response. It would be ideal if the inlucsion of the the thinking steps is optional so I can als train on datasets where I don't have thinking steps. I want to be able to train both the model shipped by Google and also this patached version: The training pipleline should work for both! During the evaluation steps I want to see the following debug output for selected samples: -The full prompt. -The exact sequence use for loss calculation (for both thinking and final reply). The se...

I run – a platform that gives small businesses their own 24/7 AI phone receptionist named “Sofie”.I need a world-class, production-grade system prompt for Retell AI that powers is not a simple chatbot or voice command handler. This is a full professional AI receptionist that must handle real customer phone calls for many different types of businesses (plumbers, HVAC technicians, dentists, salons, lawyers, chiropractors, auto repair shops, etc.) using one single global prompt must be extremely intelligent, natural, and adaptive so it works perfectly across all business types without needing separate prompts for each Requirements:Must work with Retell AI variables such as {{business_name}}, {{business_type}}, {{current_date}}, {{business_timezone}}, etc. Must automatica...

I’m looking to have a custom AI-powered chatbot built and embedded directly into my website. The primary goal is to let visitors hold natural, free-flowing conversations while the system handles user authentication in the background so returning users can be recognized and their sessions personalized. Here’s the flow I have in mind: once a visitor lands on the site, the bot greets them, authenticates them through a simple sign-in or single-sign-on step, then continues the dialogue using natural language processing to understand intent and respond intelligently. I want the language model to pick up on context over multiple turns so answers feel human rather than scripted. I already have design freedom on the site, so you’re welcome to recommend the best stack or framewor...

Hi, I’m looking to build an AI Calling Assistant for businesses. Main Goals: 1. Answer incoming calls automatically 2. Talk naturally like a human 3. Assist customers with FAQs and bookings 4. Upsell products/services during calls 5. Collect customer reviews after service 6. Transfer call to human agent if needed Features Required: - AI voice calling (inbound + outbound) - Natural conversational voice - Multi-language support (English + Indian languages preferred) - CRM integration - Call recording & transcripts - Appointment booking - Lead qualification - Upsell scripts during calls - Review collection via call/SMS/WhatsApp - Dashboard for analytics - WhatsApp/SMS follow-up automation - Ability to customize prompts/scripts Preferred Tech Stack: - OpenAI / GPT - Twilio / Ex...

I am building a call-based assistant that proactively phones users on iOS and Android, brick phone holds check-ins, and then remembers what was shared so the next conversation feels truly personal. Here’s the flow I have in mind: the agent dials the user from a mobile app, greets them naturally, explores how they are feeling, and, when appropriate, offers coping strategies or simply listens with empathy. After each call it should securely log the dialogue, extract key wellness markers, and store both the raw transcript and structured data so those insights can be surfaced in later sessions. When it starts the next call, it must be able to reference previous discussions (“Last week you mentioned difficulty sleeping—how was your rest since then?”). I’m com...

Scope of Work (SOW) Proposal LegacyLog – Experiences Knowledge Base Platform architecture to include: ● AI-powered chatbot interaction layer ● Original document/file review capabilities ● Role-Based Access Control (RBAC) ● Knowledge retrieval architecture for future AI integrations 1. Project Objective To develop a centralized internal knowledge platform (“LegacyLog”) that enables authorized users to: ● Capture organizational experiences and discussions ● Upload and preserve business knowledge ● Convert audio into searchable text ● Store and manage structured knowledge assets ● Interact with stored knowledge through an AI chatbot interface ● Enable controlled access using user roles and permissions ● Create a scalable foundation for future AI-driven business ...

Project Summary- is an AI-powered SaaS platform focused on Enterprise Performance Management (EPM), finance transformation, and reusable consulting intelligence. The platform is being designed as a scalable AI ecosystem that combines premium EPM knowledge, AI chat capabilities, contributor workflows, and future MCP (Model Context Protocol) integrations. This project is Phase 1 of the build, focusing on frontend development, UI/UX design, and creating a seamless SaaS experience. Phase 2 will be posted separately and will cover backend development, AI integration, and cloud infrastructure. Core Responsibilities The freelancer will be responsible for designing and implementing the frontend architecture and user experience for OpenEPM.ai. Main Deliverables Frontend Development: • Res...

We are looking for an AI specialist (audio & video) to upgrade our NotebookLM-generated podcasts into professional-quality videocasts. The goal is to transform automatically generated audio episodes into high-quality video content featuring realistic virtual actors who converse naturally, using cloned voices of the real actors we have selected. Project Scope: Advanced Voice Customization ============================ - Replace the default NotebookLM voices with ultra-realistic voices of the chosen actors. -Perform high-quality voice cloning from provided voice samples. - Preference for integrated or semi-integrated solutions (minimize manual export + post-processing for each episode). - Tools such as ElevenLabs, RVC, or more advanced/custom solutions are acceptable, as long as the fi...

I am looking for an experienced professional to develop a highly realistic digital avatar (digital clone / virtual human) based on my likeness. The avatar will be created from professional photographs taken in my Chroma Key studio, with full control over angles, lighting, and facial expressions. Key Requirements: Highly realistic appearance with convincing resemblance to me from multiple angles. 4K video quality. Excellent lip synchronization and natural, fluid facial expressions. Realistic voice (cloned or high-quality synthetic) with a friendly and approachable tone. Ability to speak naturally from scripts or bullet points. Desired Features: ============== - Easy to use: I want to input a script or bullet points and receive polished, ready-to-upload outputs (video, audio, or text). - W...

I need an AI agent focused on trading execution. This agent will operate on the Nse employing a day trading strategy. Key requirements: - Execute trades based on AI-generated signals - Analyze market trends in real-time - Adapt to fast-moving market conditions Ideal skills and experience: - Expertise in AI and machine learning - Strong understanding of the NSE and day trading dynamics - Proven track record in developing trading algorithms I'm looking for a developer who can create a reliable and efficient AI trading agent.

I’m a practicing physician who prefers to stay out of the spotlight, yet I still want to share trustworthy health knowledge online. I need a complete AI avatar of myself—face, voice, and personality—so I can generate original content in every format I might tackle as a solo creator: short educational videos, in-depth medical-advice articles, quick social-media posts, and anything else the platforms of tomorrow demand. The avatar must: • Look and sound convincingly like me from multiple angles and in 4K video. • Speak in a friendly, approachable tone while preserving medical accuracy. • Be easy for me to drive: I want to feed it scripts or bullet points and receive polished outputs (video, audio, or text) ready for upload. • Run on a workflow I...

I need a specialist to turn my Windows-based PC into a robust, private AI workstation. The core stack will revolve around Python and must run Llama locally through AnythingLLM while remaining flexible enough to load additional models as they mature. Key capabilities I’m after: • Decades-worth of data ingestion: PDFs, Word files, full Office docs and .eml email archives. • Context retention so I can resume a conversation or analysis session days later without re-feeding content. • Open WebUI access for convenient, browser-based interaction. • Clean restart behaviour, automated backups, and a path to scale out if I later bolt on extra GPUs or move to a small on-prem cluster. I already have the hardware; you bring the know-how to configure, script, and do...

I already have a clear plan for my funnels: now I need an engineer who can translate it into working, AI-driven automations. The focus is on three areas—marketing workflows, lead-gen pipelines, and email sequences—built around the tools I’m currently using: SocialBee for social scheduling, Brevo (formerly Sendinblue) for email delivery, Claude for language generation, and Apollo for prospect data and outreach. If additional platforms such as HubSpot, Pardot, or Marketo make sense for specific steps, I’m open to integrating them as long as they work seamlessly with the stack above. What matters most is the ability to wire these services together, apply the right AI prompts or models, and deliver clean, well-documented workflows that I can scale. Deliverables ...

Multi-LLM Governance Platform Eine Privacy-First Plattform für Unternehmen, die kontrollierten Zugriff auf mehrere moderne KI-Modelle wie ChatGPT, Claude und Gemini ermöglicht. Die Lösung kombiniert: - zentralen Unternehmenszugang - Multi-LLM-Unterstützung - Governance - Kostenkontrolle - Reporting - Rollen/Rechte - organisatorische Richtlinien Ziel ist es, Unternehmen einen offiziellen und kontrollierten Weg zur Nutzung moderner KI bereitzustellen, um Shadow AI, unkontrollierte Privatnutzung und Tool-Chaos zu reduzieren. Die Plattform ist bewusst datensparsam konzipiert: - keine Speicherung von Prompts - keine Speicherung von KI-Antworten - stateless Verarbeitung - Speicherung ausschließlich technischer Metadaten für Kosten- und Governance-Zwecke Mitarb...

FULL FREELANCER PROJECT Development of an automated forum with Discourse + AI + content control panel Project Title Development of a Discourse forum with AI automation, content generation and advanced administrative control panel ⸻ 1. Project overview I am looking for a developer or team with proven experience in Discourse, automations, APIs and integration with AI models, to create a forum-type platform initially oriented to the automotive sector. The objective is to build a functional and scalable forum based on Discourse, with the capacity to: • automatically publish threads • generate automatic responses with AI • manage multiple automated user profiles • moderate and control all activity from an admin panel • seed the forum with initial content &b...