agents001
acp-code
Coordinate a research drafter and verifier implemented with different agent frameworks.
Explore repositoryPublication in progress: 98 of 113 project repositories are uploaded and verified. The remaining repositories are being published sequentially with long pauses. Pending repository links become available after upload.
DINESH DAS · PROJECT COLLECTION
Learn the mechanism.
Build the understanding.
A working collection of AI engineering projects. Trace the source, run an experiment, examine the failure, and understand the tradeoff.
THE REPOSITORIES
Explore the collection
113 projects
memory002
agent-with-mcp-memory
Give a local agent web tools, persistent graph memory and traceable behavior.
Explore repositoryagents003
agent2agent-demo
Understand agent-to-agent communication through sine, cosine and tangent services.
Explore repositoryrag004
agentic_rag
Route questions to document retrieval and use web fallback when needed.
Explore repositoryrag005
agentic_rag_deepseek
Combine a local reasoning model with managed document parsing and search.
Explore repositorymemory006
ai-avatar-demo
Supply a conversational avatar with a memory-aware AI backend.
Explore repositorymedia007
ai-podcast-generation
Turn source material into a two-speaker podcast with cloud or local generation.
Explore repositorymedia008
ai-podcast-generator
Create article-to-podcast audio using MiniMax generation and speech.
Explore repositoryagents009
ai_news_generator
Generate sourced AI news using CrewAI and Cohere.
Explore repositorymcp010
amazon-product-analysis-server
Expose product research and comparison through an MCP server.
Explore repositorytraining011
art_mcp_rl
Train an LLM to use MCP database tools with reinforcement learning.
Explore repositoryvoice012
audio-analysis-toolkit
Transcribe audio and expose summaries, speakers and analysis via MCP.
Explore repositoryagents013
autogen-stock-analyst
Use AutoGen with a custom model client for financial analysis and code generation.
Explore repositorystructured014
biotech-agentic-analyst
Extract scientific figures and produce structured research intelligence.
Explore repositoryagents015
book-writer-flow
Generate a researched outline and chapters with a persistent workflow state.
Explore repositoryagents016
brand-monitoring
Collect brand signals across multiple platforms and synthesize reports.
Explore repositoryagents017
build-code-harness
Study a coding-agent harness with planning, tools, checkpoints and approval.
Explore repositorytraining018
Build-reasoning-model
Train a reasoning-oriented model with GRPO and explicit answer-format rewards.
Explore repositoryrag019
chat-with-audios
Retrieve answers from transcribed audio with document-style RAG.
Explore repositoryrag020
chat-with-code
Retrieve repository context for code questions and validate responses.
Explore repositoryevaluation021
code-model-comparison
Compare code-generation models on a common repository context.
Explore repositoryretrieval022
colbert-rag
Understand late interaction through ColBERT-style MaxSim scoring.
Explore repositoryrag023
Colivara-deepseek-website-RAG
Retrieve visually represented web pages and answer with a multimodal model.
Explore repositoryagents024
content_planner_flow
Plan social content from web material and schedule a draft.
Explore repositorymemory025
context-engineering-pipeline
Assemble useful context from documents, memory and structured stores under a token budget.
Explore repositorymemory026
context-engineering-workflow
Coordinate multi-source research with parsing, retrieval and long-term context.
Explore repositoryagents027
conversational-support-flow
Route support turns among order lookup, policy retrieval and live research.
Explore repositoryrag028
corrective-rag
Detect weak document evidence and recover through web search.
Explore repositorymcp029
cursor_linkup_mcp
Expose deep web search and document retrieval to an MCP client.
Explore repositorymemory030
database-memory-agent
Combine conversation history, semantic document search and tool selection in MongoDB.
Explore repositorytraining031
DeepSeek-finetuning
Adapt a distilled DeepSeek model and export it for local inference.
Explore repositoryrag032
deepseek-multimodal-RAG
Retrieve PDF pages visually and answer with DeepSeek Janus-Pro.
Explore repositoryinterfaces033
deepseek-thinking-ui
Parse streamed model traces and final answers without dropping boundary text.
Explore repositoryoperations034
deploy-agentic-rag
Serve an agentic RAG workflow through an API using LitServe.
Explore repositoryrag035
document-chat-rag
Build document-grounded question answering with local models and optional vector storage.
Explore repositoryagents036
documentation-writer-flow
Produce repository documentation through planning and writing agents.
Explore repositoryevaluation037
eval-and-observability
Build an end-to-end RAG evaluation and trace collection pipeline.
Explore repositorymcp038
eyelevel-mcp-rag
Expose complex-document ingestion and retrieval through GroundX-backed MCP tools.
Explore repositoryretrieval039
fastest-rag-milvus-groq
Explore binary-vector retrieval with Milvus and hosted generation with Groq.
Explore repositoryretrieval040
fastest-rag-stack
Combine quantized Qdrant retrieval with SambaNova-hosted generation.
Explore repositoryagents041
financial-analyst-deepseek
Expose stock analysis and generated plots through a local-model MCP workflow.
Explore repositorytraining042
finetune-studio-mcp-app
Study an MCP backend that submits and monitors Hugging Face fine-tuning jobs.
Explore repositoryrag043
firecrawl-agent
Combine corrective document retrieval with Firecrawl web fallback.
Explore repositoryagents044
flight-booking-crew
Investigate browser-assisted flight search with CrewAI tools.
Explore repositoryrag045
github-rag
Answer repository questions after GitIngest converts source to retrieval context.
Explore repositoryinterfaces046
gpt-oss-thinking-ui
Render native thinking and answer channels incrementally for local GPT-OSS.
Explore repositoryevaluation047
gpt-oss-vs-qwen3
Compare reasoning responses using parallel inference and judge metrics.
Explore repositoryrecipe048
graphiti-mcp
Study a documented integration for persistent graph memory through MCP.
Explore repositorystructured049
groundX-doc-pipeline
Process complex documents and evaluate structured extraction quality.
Explore repositorytraining050
grpo-finetuning-qwen3
Improve Qwen3 structured invoice extraction through remote GRPO training.
Explore repositoryagents051
guidelines-vs-traditional-prompt
Compare structured conversational guidelines with a traditional instruction prompt.
Explore repositoryagents052
hotel-booking-crew
Explore multi-agent hotel research with browser-backed search.
Explore repositoryoperations053
hugging-face-skills
Inspect reusable dataset, evaluation, training and publishing scripts for AI agents.
Explore repositorymedia054
imagegen-janus-pro
Study local Janus-Pro multimodal understanding and image generation.
Explore repositoryoperations055
kitops-mcp
Manage machine-learning ModelKit artifacts through an MCP tool boundary.
Explore repositorytraining056
knowledge distillation
Transfer classifier behavior from a teacher network to a smaller student.
Explore repositoryrag057
llama-4-rag
Use a hosted Llama 4 model in a local document-retrieval workflow.
Explore repositoryevaluation058
llama-4_vs_deepseek-r1
Compare Llama 4 and DeepSeek on a document-grounded task.
Explore repositorymcp059
llamaindex-mcp
Connect a local LlamaIndex agent to SQLite-backed MCP tools.
Explore repositoryinterfaces060
local-chatgpt
Build and inspect a local image-aware conversation with Llama vision.
Explore repositoryinterfaces061
local-chatgpt with DeepSeek
Understand local dialogue state and incremental DeepSeek responses.
Explore repositoryinterfaces062
local-chatgpt with Gemma 3
Study multimodal follow-ups with a compact local Gemma model.
Explore repositorymcp063
mcp-agentic-rag
Expose web-ingested retrieval through MCP using Bright Data and Qdrant.
Explore repositorymcp064
mcp-agentic-rag-firecrawl
Build an MCP retrieval service with Firecrawl ingestion.
Explore repositorymcp065
mcp-video-rag
Expose video ingestion, retrieval and clip selection through Ragie-backed MCP tools.
Explore repositoryvoice066
mcp-voice-agent
Give a LiveKit voice agent MCP-backed web and database tools.
Explore repositoryrecipe067
mindsdb-mcp
Understand federated data access through a MindsDB MCP integration.
Explore repositoryevaluation068
minimaxm2-vs-sonnet4-5-vs-kimik2-vs-gemini3
Benchmark several code-generation providers against common repository context.
Explore repositoryrag069
modernbert-rag
Explore ModernBERT embeddings in a local RAG application.
Explore repositoryagents070
motia-content-creation
Implement event-driven article-to-social-content generation.
Explore repositoryagents071
Multi-Agent-deep-researcher-mcp-windows-linux
Expose a local multi-agent research crew through MCP.
Explore repositoryrag072
multi-modal-rag
Study CLIP image retrieval and multimodal prompting through notebooks.
Explore repositoryvoice073
multilingual-meeting-notes-generator
Convert multilingual meeting audio into English notes and assigned action items.
Explore repositoryrag074
multimodal-rag-assemblyai
Unify audio and document ingestion with vector retrieval and CrewAI flows.
Explore repositoryagents075
multiplatform_deep_researcher
Research a question through platform-specific source specialists.
Explore repositoryrag076
notebook-lm-clone
Build a document-grounded notebook backend with citations, memory and audio output.
Explore repositoryevaluation077
o3-vs-claude-code
Compare repository-aware code responses and judge scoring across models.
Explore repositorymemory078
omnigraph-multi-agent
Inspect multi-agent graph consistency through provided schemas and event data.
Explore repositoryagents079
open-agent-builder
Study backend execution of AI workflows, MCP tools, guardrails and approval steps.
Explore repositoryagents080
openai-swarm-ollama
Explore local-model handoffs between lightweight agents.
Explore repositoryrecipe081
openclaw-secure-deployment
Learn host and network boundaries for an AI-agent deployment.
Explore repositoryrag082
paralegal-agent-crew
Retrieve legal-document evidence, rerank it and use research fallback.
Explore repositoryagents083
parlant-conversational-agent
Build a tool-backed loan conversation with explicit journeys and guidelines.
Explore repositorymcp084
pixeltable-mcp
Coordinate multimodal storage and computed transformations through MCP.
Explore repositoryinterfaces085
qwen3-thinking-ui
Compare thinking controls, final-answer quality and latency for Qwen3.
Explore repositoryevaluation086
qwen3_vs_deepseek-r1
Compare Qwen3 and DeepSeek on a common RAG evaluation pipeline.
Explore repositoryrag087
rag-sql-router
Route questions between document retrieval and structured SQL tools.
Explore repositoryvoice088
rag-voice-agent
Connect spoken interaction to document retrieval and speech synthesis.
Explore repositoryrag089
rag-with-dockling
Parse spreadsheets and complex documents with Docling before retrieval.
Explore repositoryvoice090
real-time-voicebot
Implement a streaming travel conversation with ASR, an LLM and TTS.
Explore repositoryrecipe091
run-jev-inside-agent-harness
Study task orchestration and credential boundaries in an agent-harness integration guide.
Explore repositoryagents092
sales-analytics-agent
Join MongoDB and HubSpot data through MindsDB for natural-language analytics.
Explore repositorymcp093
sdv-mcp
Expose local synthetic-data training, generation and statistical evaluation through MCP.
Explore repositorytraining094
siamese-network
Learn digit similarity through paired examples and contrastive loss.
Explore repositoryrag095
simple-rag-workflow
Learn the ingestion, retrieval and synthesis stages of a minimal RAG system.
Explore repositoryevaluation096
sonnet4-vs-o4
Compare code generation with shared context and DeepEval grading.
Explore repositoryevaluation097
sonnet4-vs-qwen3-coder
Evaluate Claude and Qwen3-Coder against repository-grounded tasks.
Explore repositoryagents098
stagehand x mcp-use
Connect a local agent to browser-automation tools through MCP.
Explore repositoryagents099
stock-portfolio-analysis-agent
Stream a portfolio-analysis workflow from a CrewAI backend.
Explore repositoryinterfaces100
streaming-ai-chatbot
Understand event-driven chat state and recovery through an observable local implementation.
Explore repositorytraining101
train-yolo26-object-detection
Fine-tune an object detector on a custom Roboflow dataset and track experiments.
Explore repositoryrag102
trustworthy-rag
Attach trustworthiness signals to document-grounded model output.
Explore repositorymcp103
ultimate-ai-assitant-using-mcp
Explore one agent connecting to several MCP servers.
Explore repositorymedia104
video-rag-gemini
Upload video to Gemini and ask grounded questions about its contents.
Explore repositoryagents105
web-browsing-agent
Plan and execute browser research with CrewAI and Stagehand.
Explore repositorystructured106
Website-to-API-with-FireCrawl
Convert web content into schema-constrained extraction results.
Explore repositoryagents107
Youtube-trend-analysis
Analyze channel or niche trends from collected video transcripts.
Explore repositorymemory108
zep-memory-assistant
Add persistent Zep memory to an AutoGen conversational agent.
Explore repositorymemory109
zep-observations
Study how graph conversation data becomes behavioral observations.
Explore repositoryocr110
Latex-OCR-Lama
Local Latex-OCR-Lama document recognition workspace.
Explore repositoryocr111
gemma3-ocr
Local gemma3-ocr document recognition workspace.
Explore repositoryocr112
llama-ocr
Local llama-ocr document recognition workspace.
Explore repositoryocr113
qwen-2.5VL-ocr
Local qwen-2.5VL-ocr document recognition workspace.
Explore repositoryNo projects match these filters. Try another search or reset the filters.
Source, experiments, evidence.
Each repository keeps its source attribution and includes learning material and executable CPU checks. The collection builds on open-source implementations, including AI Engineering Hub. Original model, provider, and GPU integration requires the setup documented in each project.
Read the source acknowledgments →