🤝 Top Collaborators
.png&w=640&q=75)
🔥 Own Projects
🤓 Latest Submissions
Cortex
**Cortex: Enterprise LLM Prompt Optimization & Security Gateway** Enterprise LLM workflows waste ~$34K/year on unoptimized prompts. Developers write verbose, repetitive instructions that inflate token costs across teams. There's no centralized injection attack detection. Quality is inconsistent—junior devs produce worse prompts than seniors. Cortex fixes this. It's a web-based prompt enhancement tool: paste your vibe coding prompt, get a 10x better version instantly. The system validates that all technical requirements are preserved, checks for injection attack patterns, and optimizes for token efficiency. All processing happens in-VPC—zero data leakage. **How it works:** Users enter a prompt at localhost:5000. Cortex enhances clarity and specificity using Claude 3.5 Sonnet, then validates quality preservation via dual-gate evaluation (technical parity + structural completeness). A security classifier runs semantic analysis plus regex pattern detection against 20 OWASP LLM Top 10 vectors. Returns original and enhanced versions side-by-side with metrics. **Technical pipeline:** Enhancement layer (Sonnet refines prompt structure) → Quality gate (0.75+ Q_score threshold) → Security classification (vector embeddings, cosine similarity ≥0.82) → Audit logging (tiered retention: 7d Redis, 365d Parquet, 2190d S3). **Real benchmark results (50 vibe coding prompts tested):** - 18.2% average compression - 0.75 average Q_score (all requirements intact) - 0% security flags on benign enterprise prompts - 680ms average latency (P99: 1.4s) - 100% pass rate for quality gates **Why it matters:** RAG-heavy workflows see 20% total cost reduction. Code generation workflows see 6-8%. Beyond cost: standardization. 500 developers. Every prompt quality equals senior-level engineering. Consistency across the org. **Deployment:** Flask + Anthropic API backend. Kubernetes sidecar pattern deployed in client VPC. Docker containerized. Configuration-driven (HIPAA/GDPR/SOC2 compliance policies selectable). Telemetry metadata only—no raw prompts ever hit external systems. **Competitive edge:** vs. native prompt caching (only works on 100% identical prompts—Cortex standardizes variants, boosting cache hit rates 20%→70%), vs. manual prompt engineering (automated consistency), vs. separate security vendors (single sidecar, single SLA). Cortex is real infrastructure for enterprise LLM workflows. Real data. Real compression. Real security.
27 Sep 2026
sipa-voice-gate
AI infra logs are prose. To notice that a pipeline's risk jumped between two steps, someone has to read both log lines in full and hold the first in working memory while reading the second — slow under time pressure, and structurally hostile to anyone who processes information mechanically rather than narratively. Worse: most "audit trails" just trust whatever the log says, with nothing stopping a tampered entry from looking valid. sipa-trace fixes the shape, not the content. Every pipeline event becomes a TraceCard with a locked schema (stage, action_type, risk_class, reversible, state_delta). Risk is assigned by a fixed rule table, not by asking a model to grade its own step. Two cards in a row diff field-by-field — the diff IS the alert, not a paragraph you have to parse. Every card is hash-chained, and a real verifier recomputes each entry's hash from its actual payload and compares it — so content tampering (not just chain-linkage) is caught. Core (sipa_trace/): TraceCard schema, RiskClassifier (NONE→CRITICAL), hash-chained TraceLog, diff_cards() with risk_escalated/became_irreversible flags, verify() — 29 tests green, runs with zero API keys. Instrumentation: a real MCP-server / FastAPI middleware wraps tool calls with @traced, classifying real actions (e.g. credential_access → CRITICAL) automatically as they happen. Dashboard: React + FastAPI live view polling the trace stream — trace feed, live diff panel, and a Verify badge that flips Verified → Invalid the moment a card is tampered on disk, re-checked on demand.
30 Sep 2026
Vantage
Communication's broken. No data. Just guessing. Vantage fixes it. Upload your chats—WhatsApp .txt, Discord .json, or screenshots. Get instant intelligence: Directness Score (0-10) — How clear is the communication? Detects deflection, avoidance, manipulation. No ambiguity. Effort Asymmetry Matrix — Who's actually putting in work? Tracks initiator ratio, reply delays, message length. See the imbalance. Red Flag Classifier — Breadcrumbing. Deflection. Gaslighting. Ghosting patterns. LLM-powered semantic analysis catches what humans miss. Post-Mortem Audit — Entropy shift dates. Boundary directives. Actionable recommendations: establish cadence, discontinue, or address imbalance. The Tech: FastAPI backend, Groq LLM for analysis, SQLite storage. Fast path (metrics) runs instant via SQL—no LLM cost. Slow path (red flags + audit) runs async. Raw data auto-purged after 7 days. GDPR compliant. For high-performers. Traders. Entrepreneurs. People who want data instead of vibes.
16 Sep 2026
Apex Cypher
Trading bot with the following STRATEGY: - Identifies bullish/bearish gaps between candles not filled by price action - Enters long when gap forms ABOVE VWAP - Risk: 2% account per trade, 1:3 minimum reward-to-risk - Stop loss: VWAP ± 1 ATR - Max 3 losses/day stops trading - Market orders only FEATURES: - FastAPI backend with 6 REST endpoints (equities + options) - React dashboard with live account balance, positions, PnL - Options trading support (mandatory requirement) - Paper trading integration via Alpaca API - Deployed to Render (production-ready) TECH: Python (FastAPI, Alpaca SDK), React/Vite, REST API, Alpaca Paper Trading
4 Sep 2026
MintDev
Mint is a next-generation, local-first developer tool designed to fundamentally streamline how software is built by eliminating repetitive boilerplate code. Traditional AI coding assistants often suffer from cloud latency, privacy concerns, and rigid context limits that interrupt a developer's workflow. Mint solves this by running entirely on local LLM infrastructure via Ollama, ensuring absolute data privacy, zero latency, and freedom from cloud bloat. Featuring a minimalist, Apple-inspired high-contrast aesthetic, Mint provides a distraction-free environment where developers can input natural language prompts to instantly generate clean code. Beyond initial generation, Mint introduces state-managed iterative updates, allowing creators to seamlessly modify and evolve their applications on the fly without losing context or starting from scratch. Built using Python, Flask, and a sleek Tailwind CSS interface, Mint bridges the gap between rapid prototyping and absolute local control, empowering developers to build smarter and faster.
10 Aug 2026
