APPLIED AI · DATA SYSTEMS · 25+ YEARS ON WALL STREET

AI built for defense, Hollywood, and Wall Street.

Herbert Mischel brings over 25 years on Wall Street — including 8 as Lead Data Architect for George Soros and 5 as Head of Database Architecture at SAC Capital (Point72), with engagements at Citadel, Mizuho, Bankers Trust, JP Morgan, and more. Today he builds AI systems that span SBIR-cleared DoD search-and-detect, Stable Diffusion video pipelines for Hollywood studios, and a one-shot detection framework that needs no training data.

SBIR
DoD-cleared
25+ yrs
Wall Street
13 yrs
Soros & SAC
100%
Uptime, 7 yrs

A rare combination: hedge-fund data leadership and applied AI.

Herbert Mischel brings deep, hands-on experience from both worlds — 25+ years architecting mission-critical data infrastructure across the highest levels of finance (Soros, SAC Capital / Point72, Citadel, Mizuho, JP Morgan, Bankers Trust), and a decade-plus founding and running an AI image-recognition company whose work appears in major Hollywood remasters. He has shipped AI through the Department of Defense's SBIR process for real-time battlefield intelligence.

● SBIR · DoD-cleared ● 25+ YRS · Wall Street ● FAA · certified pilot

Two engineers. Two continents. One bar.

A small, elite team with complementary backgrounds: applied AI and mission-critical data architecture out of Las Vegas, partnered with elite IDF intelligence backend engineering out of Israel.

HM

Herbert Mischel

CEO & Chief AI Officer
SBIR-CLEARED SOROS · 8 YRS SAC / POINT72 · 5 YRS FAA PILOT HARVARD EXT.

Herbert brings a rare combination of hedge-fund data leadership and applied AI innovation. 25+ years on Wall Street — including 8 as Lead Data Architect for George Soros and 5 as Head of Database Architecture at SAC Capital (Point72) — with engagements at Citadel, Mizuho, Bankers Trust, and JP Morgan, running mission-critical trading and risk infrastructure with zero downtime tolerance.

Today he architects the AI systems behind the company's defense and media work — including an SBIR-cleared search-and-detect platform for the Department of Defense, Stable Diffusion video pipelines used by Hollywood studios, and a proprietary one-shot detection framework that requires no training data.

Based: Las Vegas, NV Email: hmischel@revivalt.com
MM

Miriam "Molly" Mischel

Chief Technology Officer
IDF UNIT 81 ISRAEL SECURITY PRIZE CLOUD & DEVOPS PENN STATE

Backend engineer and systems specialist forged in Unit 81 — one of the IDF's most selective technological intelligence units. Sole architect of a mission-critical IoT monitoring and retention platform serving real-time field-device telemetry into the war room, on Azure with Docker, Kubernetes, and the ELK stack.

Contributor to a joint Israeli Air Force operation later awarded the Israel Security Prize for its impact on national security. Brings the always-on DevOps, real-time data-pipeline, and high-traffic secure-data-transfer expertise that turns research-grade AI into systems that don't fail.

Based: Herzliya, Israel Stack: Python · Ruby · Azure · K8s · ELK

The portfolio.

Production AI and data systems shipped for the U.S. Department of Defense, Hollywood studios, and the world's largest hedge funds.

◆ DoD · SBIR-cleared

AI Search-and-Detect for Battlefield Intelligence

Architected and delivered a next-generation AI search-and-detect platform that successfully passed the Department of Defense's SBIR (Small Business Innovation Research) process — qualifying it for real-time battlefield intelligence and decision support. Engineered for low latency, high reliability, and adversarial robustness, translating research-grade models into hardened operational deployments.

Process: Cleared SBIR evaluation Status: Operational
● HD Remastering · Restoration

Film Restoration & HD Remastering Suite

Going back to the original camera negative to rebuild every frame. AI-driven cleaning of dirt and scratches, color reconstruction, 4:3 → 16:9 reframing, archival-grade preservation, and recreation of destroyed special-effects shots. Plus Stable Diffusion pipelines for in-scene object replacement — branded props, ads, and assets — temporally consistent with no flicker.

Pipeline: Negative scan → AI restore → masterTech: Diffusion · LoRA · optical flow
▲ One-Shot Detection

Proprietary No-Training Detection Framework

Invented and deployed a one-shot detection framework that recognizes objects and patterns immediately — without labeled datasets and without retraining. Eliminates the data-collection bottleneck that kills most computer-vision deployments before they ship.

Training data: None requiredDeployment: Immediate
▶ Diligent Systems · 15+ yrs

AI Film Restoration & Remastering

Founded Diligent Systems and personally wrote the entire stack — UI, services, Linux/Python/Cython modules, AI training, and SQL backend. Used by Hollywood to remaster Sex and the City, Buffy the Vampire Slayer, portions of Friends, and many more. Frame-level defect removal, upscaling, color reconstruction, FX lifting, blue-screen back-plating.

Founder: Sole engineer, full stackStack: Python · OpenCV · Azure SQL
$ Soros · SAC · Citadel

Hedge-Fund Data Architecture

8 years as Lead Data Architect for George Soros and 5 years at SAC Capital (Point72), with subsequent engagements at Citadel, Mizuho, and Blue Mountain Capital. Trading and risk infrastructure with zero-downtime tolerance — 100% production uptime for 7 consecutive years at Soros.

Soros: 60+ SQL Servers, 100% uptimeSAC: 30ms IO/ps SLA
✚ Computer Vision · On-Prem

Custom Model R&D & Deployment

End-to-end model development for teams that need more than an off-the-shelf API: data pipelines, evaluation harnesses, fine-tuning, prompt engineering, LoRA-style adaptations, inference optimization for video workloads, and deployment into hard environments — including air-gapped and regulated networks.

Scope: Research → productionTooling: PyTorch · OpenCV · ONNX

Not an upscale. A rebuild.

Through Diligent Systems — the AI image-recognition company Herbert founded — every project goes back to the original camera negative and rebuilds the image from scratch using AI at every stage. Used by Hollywood to remaster Sex and the City, Buffy the Vampire Slayer, portions of Friends, and many more.

98%
Dirt & scratch removal
96%
Color accuracy
99%
16:9 composition fidelity
500+
Episodes restored
200K+
Scenes completed
Matchback Engine — Sex and the City shot match
Sex and the CityShot match · sequence analysis
Matchback Engine — Buffy location match
Buffy the Vampire SlayerLocation match · container alignment
Matchback Engine — Friends background isolation
FriendsBackground isolation · backplate recovery

"We go back to the original camera negative — then rebuild for the HD era."

Traditional restoration required armies of technicians working frame by frame — expensive, inconsistent, and slow. AI replaces that with a system that has remastered hit series watched by millions.

Three decades. Two industries. Zero downtime.

A career spanning the highest levels of finance and applied AI — from Y2K airport security to Stable Diffusion video pipelines.

Soros Fund Management · 8 years

Lead Data Architect / DBA. 60+ SQL Servers, 100% production uptime 7 years running. Spearheaded the multi-year migration to Azure SQL, Data Warehouse, and HDInsight. Built full-lifecycle iOS and web apps for real-time DB performance metrics.

SAC Capital (Point72) · 5 years

Head of Database Architecture. Eliminated latency in trade and reporting systems, consolidated SQL Servers under a 30ms IO/ps SLA, designed Fact and Dimension tables for OLAP cubes, automated BCP data distribution to downstream reporting servers.

Citadel · Mizuho · Blue Mountain

Sybase IQ → SQL Server migrations, Snowflake reporting migrations, ETL re-engineering, indexing redesigns saving terabytes, columnstore optimizations. Brought in to fix the systems other people couldn't.

Diligent Systems · Founder, 15+ yrs

Sole engineer of an AI image-recognition platform that has remastered Sex and the City, Buffy the Vampire Slayer, and portions of Friends for major studios. Recently rewritten with deep-learning modules and containerized for Azure SQL.

U.S. Department of Defense · SBIR

Architected and delivered an AI search-and-detect platform that successfully passed the Department of Defense's SBIR (Small Business Innovation Research) evaluation, qualifying it for real-time battlefield intelligence and decision support.

Port Authority NY/NJ · Y2K

Lead DBA for the database side of the Y2K conversion for air-side security at Newark, JFK, and LaGuardia. Implemented high-security logging and monitoring; interfaced with local and federal law enforcement on national security testing.

What goes into every system.

The same rigor that cleared SIBR applies to every project — whether the adversary is enemy optics, denied claims, a studio deadline, or a regulator.

Data & evaluation pipelines

Ingest, label, version, and measure. No model ships without a harness that tells you when it regresses.

Compliance & validation

SIBR, HIPAA-aware workflows, auditable model provenance. Built to survive the review that comes after the demo.

Edge & real-time inference

Quantized, distilled, and optimized for the hardware you actually deploy on — not a reference GPU in a datacenter.

Multi-modal systems

Vision, language, audio, structured data — fused into systems that make decisions a human operator trusts.

Have a hard problem?

If you're looking for AI that has to work the first time — in the field, in the clinic, on set, or inside a regulated network — let's talk.

hmischel@revivalt.com →