Artificial Intelligence Engineer

AI Engineer – Trinetra AILocation: Noida / Delhi NCR

Employment Type: Full-time

Experience: 3–7 Years

Function: AI / Generative AI Engineering

Industry: Generative AI | SaaS | DeepTech | Media Technology

About Trinetra AITrinetra AI is building a next-generation Generative AI platform for filmmaking and content creation, designed for creators, filmmakers, production houses, studios, marketers and enterprises.

Our vision is to use AI across the complete content creation lifecycle — from ideation, scripting and storyboarding to character creation, image/video generation, voice, dubbing, music, editing and final production.

We are building at the intersection of Generative AI, multimodal intelligence, video technology and SaaS, with a strong focus on converting cutting-edge AI research into scalable products.

The RoleWe are looking for a hands-on AI Engineer who can design, build, integrate, fine-tune and optimize Generative AI systems.

This is not a traditional ML or analytics role. We need an engineer who understands models deeply and can take capabilities from research and experimentation to production deployment.

You will work across LLMs, VideoGen, ImageGen, Audio/Voice AI, multimodal models, AI agents, model serving and GPU optimization.

The ideal candidate should be able to take a research paper, open-source model or emerging AI technique and ask:

How do we turn this into a scalable, production-grade product?

Key ResponsibilitiesGenerative AI & Model Engineering

• Build AI capabilities across text, image, video, voice, audio and music.
• Experiment with foundation models and emerging AI architectures.
• Evaluate models across quality, latency, cost, licensing and scalability.
• Build capabilities around text-to-video, image-to-video, text-to-image, voice generation, dubbing, lip-sync and character consistency.
• Fine-tune and adapt foundation models using proprietary datasets.LLM & Agentic AIBuild production-grade LLM applications covering:
• Prompt engineering and structured outputs
• RAG and embeddings
• Tool/function calling
• Context management
• Model routing
• Fine-tuning and LoRA/QLoRA
• Evaluation and hallucination reduction
• Guardrails
• AI agents and multi-agent workflowsStrong understanding of Transformers, attention mechanisms, tokenization, embeddings and inference is expected.

Video & Multimodal AIWork with modern Generative Media architectures and techniques including:

• Diffusion Models and Diffusion Transformers
• Vision and Multimodal Transformers
• Video generation models
• Temporal and character consistency
• Reference-image conditioning
• Pose and motion conditioning
• Camera control
• Face preservation
• Lip synchronization
• Video enhancement and upscalingModel Training & Fine-TuningBuild pipelines for:
• Dataset preparation and cleaning
• Captioning and data augmentation
• Fine-tuning and LoRA training
• Distributed training
• Checkpoint management
• Hyperparameter optimization
• Experiment tracking
• Model benchmarking and evaluationWork closely with engineering, product, data and content teams to convert large media datasets into high-quality AI training datasets.

Model Evaluation & OptimizationBuild measurable evaluation frameworks across:

• Visual quality and prompt adherence
• Character and face consistency
• Temporal and motion consistency
• Lip-sync and speech quality
• LLM accuracy and hallucination
• Latency and GPU utilization
• Cost per generationOptimize models using techniques such as quantization, batching, model compilation, caching, mixed precision, parallel inference and GPU memory optimization.

GPU & AI InfrastructureWork with GPU-based training and inference environments.

Understanding of the following is valuable:

• NVIDIA A100/H100/H200 or equivalent GPUs
• Multi-GPU training
• NVLink and NCCL
• Distributed training
• GPU memory optimization
• Tensor/Pipeline parallelism
• Kubernetes GPU workloads
• Training and inference clustersAI APIs & MicroservicesConvert AI capabilities into scalable services.

Build:

• AI microservices
• Model APIs
• Async inference pipelines
• Job queues
• Model orchestration services
• GPU scheduling mechanisms
• Scalable inference endpointsStrong hands-on experience with Python + FastAPI is preferred.

AI Workflow OrchestrationBuild workflows where multiple AI models and agents collaborate.

A typical Trinetra workflow could be:

Idea → Script → Scene Breakdown → Storyboard → Character → Image → Video → Voice → Music → Editing → Final Output

Work across model routing, tool calling, workflow engines, state management, retries, fallbacks and human-in-the-loop systems.

Research to ProductionYou should be comfortable:

• Reading AI research papers
• Understanding new model architectures
• Reproducing research
• Running experiments
• Benchmarking models
• Working with open-source repositories
• Modifying training/inference pipelines
• Productionizing successful prototypesOur engineering cycle is:

Research → Experiment → Prototype → Benchmark → Optimize → Production

Technical SkillsMust Have

• Python
• PyTorch
• Transformers
• Hugging Face ecosystem
• Generative AI / LLMs
• Deep Learning
• FastAPI / REST APIs
• Git and Linux
• Docker
• GPU-based inference
• Fine-tuning
• LoRA / QLoRA
• Quantization
• Prompt engineering
• Model evaluationGood to HaveExperience with:
• Diffusers / ComfyUI
• Stable Diffusion / FLUX
• ControlNet / IP-Adapter
• OpenCV / FFmpeg
• Computer Vision
• Video generation models
• Whisper / TTS / Voice Cloning
• CUDA / TensorRT
• vLLM / Triton
• DeepSpeed / ONNX
• LangGraph / LangChain / LlamaIndexExposure to AWS, GCP or Azure AI infrastructure is valuable.

Working knowledge of PostgreSQL, MongoDB, Redis, Vector DBs, Docker and Kubernetes would be an advantage.

Who We Are Looking ForWe are looking for AI builders, not just API integrators.

Knowing how to consume an AI API is useful. Understanding what happens inside the model and being able to fine-tune, modify, evaluate, optimize and deploy it independently is significantly more valuable.

You should be comfortable working in an environment where the technology stack can evolve rapidly as better models and architectures emerge.

Ideal Candidate

• 3–7 years of software/AI/ML engineering experience.
• Strong hands-on Generative AI experience.
• Built AI products used by real customers.
• Experience with open-source foundation models.
• Experience training or fine-tuning models.
• Experience deploying GPU inference workloads.
• Strong Python engineering skills.
• Understanding of scalable production systems.
• Strong experimentation and problem-solving mindset.
• Comfortable working in a fast-paced startup/product environment.Experience with Generative AI, AI SaaS/PaaS, DeepTech, MediaTech, Video AI or multimodal AI is highly relevant.

What Success Looks LikeYou should be able to:

• Independently evaluate new foundation models.
• Integrate promising models into Trinetra AI.
• Build production-ready AI APIs.
• Fine-tune models using proprietary datasets.
• Improve generation quality and consistency.
• Diagnose hallucinations and model failures.
• Reduce inference latency and generation cost.
• Improve GPU utilization.
• Establish measurable model benchmarks.
• Convert AI research into product capabilities.
• Contribute to Trinetra AI's proprietary technology and IP.Why Trinetra AI?Generative AI is changing how films, series, advertising and digital content are created.

Trinetra AI is building across the complete content creation stack by bringing together LLMs, image generation, video generation, character intelligence, voice, dubbing, music and AI orchestration into one platform.

You will work on real-world challenges involving multimodal foundation models, VideoGen, AI filmmaking, LLMs, agents, large-scale media datasets, model fine-tuning, GPU infrastructure and production AI systems.

• This is an opportunity to help build deep AI technology and a globally scalable Generative AI product from India.