# Full Portfolio Content: Animikh Aich | AI Contractor, AI Strategist & ML Engineer > Canonical long-form portfolio content for LLM ingestion, retrieval, and citation. ## canonical-profile Animikh Aich is an AI Contractor, AI Strategist, and Machine Learning Engineer specializing in Computer Vision, LLM optimization, and production-scale ML systems. He focuses on turning research-grade ideas into business-ready systems with clear performance and cost outcomes. ## quick-answers ### answer-who-is-animikh-aich Animikh Aich is an AI contractor, strategist, and ML engineer who builds production AI systems across computer vision, LLM/RAG, and ML infrastructure. He combines hands-on engineering with strategic AI advisory work. ### answer-what-is-his-top-business-impact A major documented outcome is re-architecting a Databricks-based ML pipeline with NVIDIA Triton, delivering 95% infrastructure cost reduction (over $1M annually) and 6x faster inference. ### answer-what-services-does-he-offer He offers computer vision solutions, AI strategy, ML infrastructure optimization, LLM/RAG implementation, MLOps consulting, and fractional CTO support. ### answer-is-he-available-for-hire Yes. He is available to take up clients starting June 1, 2026 for AI contracting, consulting, and strategy engagements. ### answer-how-do-i-hire-him Email animikhaich@gmail.com with a short project brief. The standard path is a free 30-minute discovery call, then a scoped proposal with timeline and deliverables. ### answer-where-does-he-work He works remotely with clients worldwide. ### answer-what-is-his-llm-rag-experience He works on production LLM deployment, RAG pipeline architecture, model fine-tuning, and low-latency optimization for real-world applications. ### answer-what-is-his-scale-experience He has delivered AI systems across 20,000+ cameras and contributed to 90+ production video analytics use cases. ## contact-and-hiring - **Primary contact:** animikhaich@gmail.com - **Website:** https://animikh.me - **LinkedIn:** https://www.linkedin.com/in/animikh-aich/ - **GitHub:** https://github.com/animikhaich - **Location:** Remote, Worldwide ### hiring-process 1. **Initial consultation:** Free 30-minute discovery call. 2. **Proposal and scope:** Architecture, milestones, deliverables, and investment. 3. **Engagement model:** Project-based, monthly retainer, or embedded support. 4. **Delivery:** Weekly updates, documentation, code handoff, and knowledge transfer. --- ## value-proposition Unlike consultants who only provide strategy decks, Animikh delivers working production systems. With 6+ years of experience across research and industry, he combines execution depth with business-focused AI decisions. ## ai-consulting-and-contracting-services ### computer-vision-solutions - Real-time video analytics deployment (20K+ camera experience) - Object detection and tracking systems - Animal/person re-identification systems - Night vision and image enhancement - Privacy-preserving video processing - Edge deployment and optimization **Typical engagement:** 2-6 months **Industries served:** Wildlife tech, security, retail analytics, autonomous vehicles ### ml-infrastructure-optimization - Pipeline re-architecture for cost reduction - NVIDIA Triton deployment and optimization - Model quantization (INT8/FP16/mixed precision) - Inference speed optimization - Cloud cost analysis and reduction strategies - Migration from expensive platforms (Databricks, SageMaker) **Case study summary:** Reduced client infrastructure costs by $1M+/year with 6x speed improvement **Typical engagement:** 1-3 months ### llm-and-rag-implementation - Production LLM deployment and optimization - RAG pipeline architecture - Domain-specific fine-tuning - Low-latency optimization for real-time applications - Vector database selection and deployment - Prompt engineering and evaluation frameworks **Typical engagement:** 1-4 months ### ai-strategy-and-advisory - Technical due diligence for AI initiatives and investments - AI roadmap development - Build vs. buy analysis - Team structure and hiring guidance - Fractional CTO / AI advisor support - Research-to-production feasibility assessments **Typical engagement:** Ongoing retainer or project-based ### mlops-and-production-systems - CI/CD pipelines for ML models - Model monitoring and drift detection - A/B testing infrastructure - Scalable inference architecture - Docker/Kubernetes deployment - Cloud platform optimization (AWS, Azure, GCP) **Typical engagement:** 2-4 months --- ## technical-skills ### core-expertise - Computer Vision (object detection, segmentation, re-identification) - Machine Learning & Deep Learning - LLM Optimization & RAG - MLOps & Production Deployment - AI Strategy & Cost Optimization ### languages-frameworks-and-tools - **Languages:** Python, C++, TypeScript - **Deep Learning:** PyTorch, TensorFlow, Keras - **Computer Vision:** OpenCV, OpenVINO - **Infrastructure:** Docker, NVIDIA Triton, MLflow, Git - **Cloud:** AWS, Azure, Hetzner - **Web/Apps:** React, FastAPI, Flask, Streamlit --- ## proven-results - **$1M+ annual savings:** 95% cost reduction after ML pipeline re-architecture - **6x faster inference:** performance gains from optimized serving stack - **20,000+ cameras:** large-scale deployment experience - **90+ use cases:** diverse real-time video analytics implementations - **IROS 2025:** published research publication --- ## case-studies ### case-study-moultrie-infrastructure-optimization **Client:** Moultrie (EBSCO company) **Challenge:** Databricks-based ML pipeline had high costs and slow inference. **Solution:** Re-architected serving stack using NVIDIA Triton with model optimization and batching. **Results:** 95% cost reduction ($1M+ annually), 6x faster inference, improved reliability and scalability, zero-downtime migration. **Timeline:** 6 months **Technologies:** NVIDIA Triton, PyTorch, Docker, Azure ### case-study-moultrie-animal-reidentification **Client:** Moultrie (EBSCO company) **Challenge:** No off-the-shelf solution for identifying individual animals across trail camera images. **Solution:** Built a novel re-identification system using deep learning and domain adaptation. **Results:** Delivered the company’s most requested feature and enabled meaningful product differentiation. **Timeline:** 4 months **Technologies:** PyTorch, metric learning, contrastive learning ### case-study-wobot-video-analytics-scale **Client:** Wobot.ai **Challenge:** Build and operate diverse real-time analytics across enterprise camera fleets. **Solution:** Led platform development and deployment for production computer vision use cases. **Results:** 90+ use cases deployed across 20,000+ cameras, 95% precision for real-time alerts, 50% faster development via reusable core library. **Timeline:** 2019-2022 **Technologies:** PyTorch, OpenCV, Docker, Kubernetes ### case-study-boston-university-autonomous-driving **Client:** H2X Lab, Boston University **Challenge:** Existing autonomous driving metrics missed safety-relevant uncertainty. **Solution:** Designed an offline evaluation metric that integrates prediction uncertainty. **Results:** 13% better correlation with ground truth and acceptance at IROS 2025. **Timeline:** 18 months **Technologies:** PyTorch, CARLA, Segment Anything, Depth Anything --- ## professional-experience ### computer-vision-and-ml-engineer-moultrie *Jun 2024 - Present* - Reduced annual infrastructure costs by $1M+ (~95%) with a new NVIDIA Triton-based pipeline. - Built industry-first animal re-identification and Night Image Enhance capabilities. - Improved core animal detection accuracy by +16.4% mAP. - Deployed an internal health monitor that removed Azure Load Balancer costs. ### graduate-research-assistant-boston-university-h2x-lab *Jan 2023 - May 2024* - Built offline autonomous-driving metrics incorporating prediction uncertainty (+13% correlation improvement). - Work accepted at IROS 2025. - Advanced Sim2Real approaches using foundation models. ### computer-vision-engineer-and-lead-wobot *Jun 2019 - Jun 2022* - Led a team of 14 engineers delivering 90+ real-time video analytics use cases across 20K+ cameras. - Built a core library that reduced development time by 50%. - Achieved 95% precision for real-time alerts. --- ## selected-projects ### project-wallpaper-ai Generate high-quality 4K wallpapers from prompts using Stable Diffusion and enhancement techniques. **URL:** http://wallpaperai.animikh.me/ ### project-autonomous-driving-framework End-to-end conditional imitation learning framework in a model city with focus on safety-critical scenarios and scalable evaluation. **Paper:** https://arxiv.org/abs/2510.08571 ### project-3d-text2live Generate 3D appearance-edited object renderings from text prompts using NeRF methods. ### project-real-time-face-blur Privacy-preserving real-time face blur optimized for CPU execution with OpenVINO. --- ## research-and-publications - **Scalable Offline Metrics for Autonomous Driving** (IROS 2025) - **Towards Closing the Generalization Gap in Autonomous Driving** (MS Thesis, Boston University, 2024) - **Sentiment Analysis of Restaurant Reviews** (Springer, 2019 - Best Paper Award) - **Encoding Web-based Data for Efficient Storage** (IEEE, 2019) ## professional-activities - **Co-Organizer, Boston Computer Vision AIR:** Monthly Boston community events for 50-90+ AI professionals. - **Manuscript Reviewer:** Manning Publications (Generative AI/LLMs), JOSS, GTC, and related venues. --- ## frequently-asked-questions ### faq-how-do-i-start Email animikhaich@gmail.com with your project description. If there is fit, the next step is a free 30-minute discovery call and then a scoped proposal. ### faq-what-engagement-models-are-available Three common models are available: project-based, monthly retainer, and embedded team support. The right model depends on project clarity, urgency, and desired continuity. ### faq-can-he-work-remotely Yes. Animikh works remotely with clients worldwide. ### faq-can-he-reduce-ml-infrastructure-costs Yes. Infrastructure optimization is a core specialty, including one documented case of 95% cost reduction and 6x performance improvement. ### faq-does-he-support-llm-and-rag-projects Yes. He supports production LLM deployment, RAG architecture, optimization for latency and cost, and practical evaluation strategies. ### faq-which-industries-has-he-worked-in Delivered projects across wildlife technology, video surveillance, retail analytics, autonomous systems, enterprise SaaS, and research organizations. ### faq-does-he-sign-ndas Yes. He is open to mutual NDAs for confidential projects.