CFO-Copilot
Autonomous financial agent with SmolAgent, RAG over ERP/PostgreSQL, Qdrant, Redis caching, and Chronos-2/N-Hits/Prophet forecasting.
Hey There, This is
Shashi Ranjan
AI & ML Engineer · Bengaluru, Karnataka
And I'm an
About Me
SYSTEM BRIEF
Building AI & ML systems that ship in production.
AI & ML Engineer based in Bengaluru, Karnataka, currently building autonomous agent systems at Thrivv AI (Remote, Dubai). I specialize in machine learning engineering (feature pipelines, model training, MLflow, drift monitoring), MCP-powered agents, hybrid RAG, LLM fine-tuning, eval-driven LLMOps, and production inference pipelines — from multimodal interview systems to financial copilots and document forensics engines.
CAREER LOG
Previously at Vacanzi as AI Engineer and at Rubixe AI Solution as a Data Scientist Consultant, I've delivered classical ML models, forecasting systems, RAG chatbots, computer vision pipelines, and client analytics solutions. I work across PyTorch, Scikit-Learn, XGBoost, LightGBM, MLflow, Airflow, LangChain, LangGraph, SmolAgent, MCP, vLLM, FastAPI, Qdrant, Redis, PostgreSQL, and AWS — with a focus on evaluated, guarded, low-latency production AI & ML systems.
ML Engineering
Feature pipelines, XGBoost/LightGBM, MLflow registry, drift monitoring, and batch/online inference.
Agentic AI
MCP tool agents, hybrid RAG, and LoRA-tuned models built for real workloads.
Production Ready
Evals, guardrails, and observable ML/LLM pipelines aligned to business outcomes.
WHY WORK WITH ME
I combine deep technical expertise in machine learning engineering, agentic AI, hybrid RAG, and production LLMOps with clear communication — translating complex engineering into business outcomes. I invest time upfront to understand the problem, then build evaluated, guarded, scalable systems that perform in production.
PyTorchXGBoostMLflowAirflow
LangGraphMCPQdrantAWS
Experience
AI Engineer
Thrivv AIProject 1: CFO-Copilot (Autonomous Financial Agent)
- Built a secure locally-hosted LLM agent using SmolAgent with MCP tool integrations.
- Developed a RAG pipeline querying ERP systems and PostgreSQL via Qdrant vector search and Cohere reranking.
- Implemented Redis caching to accelerate financial data retrieval and reduce query latency.
- Integrated Chronos-2, N-Hits, and Prophet forecasting models for live banking predictions.
Project 2: Document Tampering Detection System
- Developed digital image forensics using DCT, FFT, and Error Level Analysis (ELA).
- Implemented EXIF/XMP metadata validation for timestamp and editing-software detection.
- Built PDF structural analysis for font embedding and manipulation artifacts.
- Designed Llama 3 explainability engine for automated authenticity reports.
AI Engineer
VacanziProject 1: AI Interview Simulator
- Built a real-time multimodal AI system for autonomous video interviews.
- Designed agent workflows with LangChain and GPT for contextual follow-up questions.
- Implemented CV pipelines with OpenCV and TensorFlow for eye movement and facial behavior.
- Integrated audio analysis for voice modulation and communication patterns.
Project 2: Agentic Assistance Chatbot
- Developed a voice-enabled AI assistant for job postings and recruiter data.
- Built RAG with open-source LLMs and LangChain for real-time platform queries.
- Implemented TTS/STT pipelines for full conversational voice interaction.
- Connected MongoDB for persistent profiles and live recruiter job data.
Data Scientist Consultant (Intern)
Rubixe AI Solution Pvt. Ltd.- Built ML models including ANNs achieving 92% prediction accuracy.
- Delivered analytical dashboards and reports for 10+ clients, improving decision speed by 20%.
- Automated SQL workflows and data pipelines, reducing manual effort by 40%.
My Skills
Languages & DSA
PythonSQLDSABashTypeScriptJavaC++
ML Engineering
Feature EngineeringMLflowModel RegistryFeature Store
AirflowSparkKafkaRay
Drift MonitoringA/B TestingContinuous TrainingBatch Inference
AI / ML
PyTorchTensorFlowScikit-LearnXGBoost
LightGBMOptunaHugging FaceOpenCVYOLO
LLM & Agents
LangChainLangGraphSmolAgentMCPCrewAIAutoGenStructured Outputs
RAG & Retrieval
RAGCRAGHybrid SearchEmbeddingsQdrantpgvectorCohere Rerank
Fine-Tuning & LLMOps
LoRAQLoRAPEFTRagasDeepEvalGuardrailsLangSmithW&B
Inference & Serving
vLLMONNXAWQGPTQbitsandbytesStreaming
Backend & Cloud
FastAPIDockerKubernetesAWSCI/CDRedisGit
Data & Model APIs
PostgreSQLMongoDBOpenAIAnthropicHugging Face HubBedrock
Multimodal
WhisperCoqui XTTSVLMsOpenCVMediaPipeLibrosa
CERTIFICATIONS
-
Certified Data Analyst Datamites Global Institute · Bengaluru
-
Certified Data Analyst IABAC · International Business Analytics
INDEPENDENT PROJECTS
-
ML Engineering & Forecasting ANNs · XGBoost / LightGBM · Prophet · Optuna · Scikit-Learn
-
Custom TTS Fine-Tuning Coqui XTTS v2 · ONNX · Whisper · Librosa
-
LLM Fine-Tuning — Mistral 7B PEFT · LoRA · QLoRA · Hugging Face · PyTorch
-
Data Analysis Projects IMDB SQL analysis · HR analytics (Power BI / Tableau)
My Projects
LLM Fine-Tuning (Mistral 7B)
Fine-tuned Mistral 7B v0.3 with PEFT (LoRA, QLoRA), custom preprocessing pipelines, and domain-specific evaluation.
Custom TTS Fine-Tuning
Coqui XTTS v2 fine-tuned on Indian accent data with Librosa/Whisper preprocessing and ONNX inference optimization.
Document Tampering Detection
Digital forensics engine using DCT, FFT, ELA, EXIF validation, PDF analysis, and Llama 3 explainability reports.
AI Interview Simulator
Real-time multimodal video interviews with LangChain agents, OpenCV/TensorFlow CV, and audio communication analysis.
Agentic Assistance Chatbot
Voice-enabled RAG chatbot with TTS/STT pipelines and MongoDB for live recruiter job data.
IMDB & HR Data Analysis
SQL analytics on IMDB movies and HR dashboards in Power BI / Tableau for decision-ready insights.
My Services
Machine Learning Engineering
Design end-to-end ML systems — feature engineering, XGBoost / LightGBM / Scikit-Learn / PyTorch models, Optuna tuning, offline evaluation, and online or batch inference at production scale.
ML Pipelines & Platform MLOps
Build continuous training pipelines with Airflow, Spark, Kafka, Ray, MLflow model registry, feature stores, drift monitoring, A/B tests, and CI/CD for reliable model rollout.
Agentic AI Systems
Design and deploy autonomous LLM agents with MCP tools, SmolAgent, LangGraph, structured outputs, multi-step reasoning, and secure local or cloud hosting for enterprise workflows.
RAG & Retrieval Systems
Build production RAG and CRAG with hybrid BM25 + dense search, embeddings design, Qdrant or pgvector, Cohere reranking, Redis caching, and live ERP / PostgreSQL / MongoDB sources.
LLM Fine-Tuning & Alignment
Fine-tune open-source LLMs (Mistral, Llama) with PEFT, LoRA, and QLoRA on Hugging Face — including preference-style alignment workflows and domain adapters.
LLMOps, Evals & Observability
Ship eval-driven AI with Ragas / DeepEval harnesses, LangSmith tracing, W&B experiment tracking, regression suites for agents, and cost / latency monitoring.
Inference Serving & Optimization
Serve models at scale with vLLM, ONNX, AWQ / GPTQ / bitsandbytes quantization, streaming APIs, FastAPI, Docker, Kubernetes, and AWS GPU deployments.
Financial AI & Forecasting
Deliver CFO copilots and financial agents with ERP integration, banking data RAG, and time-series forecasting using Chronos-2, N-Hits, and Prophet.
Computer Vision & Multimodal AI
Build real-time CV and VLM pipelines with OpenCV, YOLO, MediaPipe, TensorFlow, and vision-language models for video interviews, behavior analysis, and document vision.
Voice AI (TTS & STT)
Build conversational voice assistants with Whisper STT, Coqui XTTS custom accent fine-tuning, Librosa audio pipelines, and low-latency duplex UX.
Document Intelligence & Forensics
Detect document tampering using DCT, FFT, ELA, EXIF/XMP validation, PDF structural analysis, and Llama-powered explainability reports.
Guardrails & AI Safety
Add production guardrails — moderation, PII redaction, prompt-injection defenses, policy filters, and audit-friendly logging for enterprise AI assistants.
LLM API & Platform Integration
Integrate OpenAI, Anthropic, Hugging Face, and AWS Bedrock into products with robust prompting, structured outputs, streaming, guardrails, and cost-optimized architectures.
Education
| Institution | Degree | Year | Grade / Status |
|---|---|---|---|
| Indian Institute of Technology Patna | Master of Computer Application (MCA) | Pursuing | CPI: 9.35/10 · SPI: 9.35/10 |
| Bengaluru City University | Bachelor of Computer Application (BCA) | 2023 | First Class |
Contact Me
Hire an AI & ML engineer who ships.
Available for freelance and contract AI / ML engineering — ML pipelines, agentic systems, hybrid RAG, LLM fine-tuning, evals, guardrails, and production inference. Based in Bengaluru — open to remote worldwide.
Available for hire
Name
Shashi Ranjan Kumar Singh
Location
Bengaluru, Karnataka · Remote Worldwide
Phone / WhatsApp
+91 6204637985
Languages
English & Hindi
AI ASSISTANT
Have a question? Ask to AI
Chat about Shashi's experience, skills, projects, services, or how to collaborate on AI engineering work.