DATADOCTORZ is a Chennai-based technology enterprise established in 2008 with offices in Ambattur and Nungambakkam, Tamil Nadu. We deliver Industrial AI, Embedded AI, Machine Vision, Manufacturing Execution Systems, and Legacy Data Recovery services across India.
Professional-grade data recovery for HDD, SSD, RAID, NAS, and flash media. Forensic-level extraction from mechanically failed media. Clean-room capability, in-lab and on-site options, confidential handling with non-disclosure assurance. Operational since 1998.
Tailored MES platform with real-time production visibility integrating PLCs, SCADA, ERP, and MIS. OEE dashboards, quality management, traceability, reject analysis. Multi-brand PLC support including Siemens, Allen-Bradley, Delta, and Mitsubishi. Delivered as SaaS — low capex, rapid deployment.
Connects existing machines — CNC, PLCs, conveyors, motors, sensors — streaming live data to desktop, database, or cloud ERP/MIS. Supports Modbus RTU/TCP, OPC-UA, MQTT, RS-232/485, Ethernet/IP, Profibus. Edge gateway with local buffering ensures zero data loss on network drops.
Enterprise-grade Face Recognition and Visitor Management system. Uses InsightFace/ArcFace deep learning achieving ±99% recognition accuracy at 30 FPS. Features self-learning engine with adaptive per-person threshold tuning, MAC-address-bound licensing, offline operation capability, and standalone executable deployment requiring no Python runtime on target machines. MySQL/MariaDB backend with remote VPS support.
Advanced industrial machine vision and image synthesis platform for quality inspection, defect detection, and intelligent imaging in manufacturing environments. Supports multi-camera synchronised capture, conveyor, robotic, and CNC vision cell integration, and AI training dataset generation through image synthesis and augmentation.
Flagship Industrial AI Code Generation Platform using multi-LLM models to produce production-ready code across PLCs, HMIs, SCADA, Robotics, and CNC. Supports 13 PLC brands: Siemens, Allen-Bradley, Mitsubishi, Omron, Delta, Beckhoff, Unitronics, Schneider, ABB, Fatek, Fuji, LS, Keyence. Supports 11 robot brands: Fanuc, KUKA, ABB, Yaskawa, Kawasaki, UR, Denso, Nachi, Panasonic, Stäubli, Epson. Native file format downloads — ladder, ST, FBD, SFC — for direct IDE import. IEC 61131-3 compliant.
BASIC tier: single user, core PLC and HMI generation. PRO tier: multi-user, full library including PLC, Robot, HMI, and SCADA with native exports. ENTERPRISE tier: unlimited users, on-premise deployment, custom integration, dedicated SLA, and white-label OEM packaging. Available as SaaS or on-premise.
SoftGPU is DATADOCTORZ's patent-pending portable warp-compute abstraction layer written in pure C. Indian Provisional Patent Application No. 202641043907 filed April 6, 2026. PCT filing deadline April 6, 2027. Benchmark: 377 million operations per second on AMD Ryzen 7 5700G using 8 warps and 256K floats.
Self-adjusting 4-tier backend architecture: Vulkan/RADV for x86 GPU offload, AVX2/FMA3 for x86 CPU, ARM NEON for Cortex-A, and Scalar C for bare-metal MCUs. Warp-to-RTOS-task mapping enabling GPU-style workgroups on bare-metal RTOS — patent pending. IEC 61131-3 PLC co-execution on same silicon — industry first, patent pending. Watchdog-safe rollback ensures LLM inference never preempts PLC scan cycle. Zero dynamic allocation post-init — safe for safety-critical deployments.
ESP32-S3 with Xtensa LX7 and PSRAM for edge inference. STM32F407 Cortex-M4 for bare-metal RTOS warp with MPU sandbox. ARM Cortex-A on Orange Pi 5 and Raspberry Pi with NEON SIMD and NVMe tiering for large models. AMD Ryzen and Intel x86-64 with AVX2 and FMA3 vectorised warps. Vega 8 and Vulkan iGPU with RADV backend. Same binary adapts from 240 MHz MCU to 4.6 GHz x86 workstation.
Warp-parallel Opus, G.711, and G.729 codec execution on MCU or x86. Sub-5 millisecond mouth-to-ear latency on ARM Cortex-M4. Echo cancellation and noise suppression as shader-mapped DSP kernels. Integrates with lwIP stack on FreeRTOS. No external DSP chip or dedicated audio SoC required.
Full software IP Private Branch Exchange with SIP and RTP switching, conferencing, and IVR on embedded Linux or bare-metal ARM. Multi-channel SIP session handling with warp-parallel RTP mixing. Conference bridge using parallel warp reduction kernels. Scales from 4-channel embedded PBX to 128-channel rack unit on x86. Deterministic scheduling ensures no audio glitch under PLC co-load.
32 to 64 channel parallel DSP including EQ, dynamics, and convolution reverb as warp kernels. Dolby Atmos-style object audio spatial rendering on x86 AVX2 warps. Sub-1 millisecond DSP round-trip latency with zero-copy warp pipeline. Room correction FIR filters. Runs on fanless embedded x86 or ARM with no PCIe DSP card required. Compatible with ALSA, JACK, and custom bare-metal audio HAL.
Real-time fusion of IMU, LiDAR, camera, pressure, and environmental sensor data using warp-parallel Kalman, Mahony, and Madgwick filters. Deterministic 1 kHz fusion rate co-existing with IEC 61131-3 PLC task. Neural pose estimation quantised to Q4 for MCU execution. Used in AGV navigation, drone stabilisation, and industrial robot arms.
Hard real-time AI inference with warp-to-RTOS-task mapping giving every neural network layer a provable deadline. WCET (Worst-Case Execution Time) provable per warp on Cortex-M4. Quantised Q4 and Q8 TinyML neural nets execute as deterministic shader passes. Suitable for ISO 26262 ASIL-B and IEC 61508 SIL-2 system integration. First soft-real-time AI co-processor implemented in pure C without a GPU runtime.
Quantised large language model inference including Llama, Phi, and Mistral entirely on edge hardware. Q4_K_M and Q8 llama.cpp-compatible inference on ARM and x86. NVMe and zRAM memory tiering enables 7 billion parameter models on 4 GB DRAM targets. Runs fully offline with no cloud, no GPU, no internet dependency. SoftGPU scheduler prevents LLM from starving real-time control tasks.
World's first architecture enabling IEC 61131-3 PLC ladder and ST logic to co-execute with neural network inference on the same silicon. PLC scan cycle of 1 to 10 milliseconds is never missed. AI inference fills idle warp slots between PLC scan windows. Demonstrated on STM32F407 Cortex-M4 and AMD Ryzen 7 5700G. Enables self-programming PLCs that learn from process data on-chip. Patent Pending: Indian Provisional Application No. 202641043907.
SoftGPU enables embedded systems to observe their own runtime behaviour, infer corrective parameters using on-chip neural inference, and rewrite their own control logic down to raw HEX firmware. No programming language barrier: output spans IEC 61131-3 Ladder, ST, FBD, C, Assembly, Intel HEX, and Motorola SREC firmware images ready for direct flash. The observe, infer, rewrite, compile, and flash loop completes on-device in under 500 milliseconds with no PC, no IDE, and no programmer dongle required. Watchdog-gated: newly generated HEX is sandbox-executed in a shadow warp and CRC-verified before live flash. Unsafe images are silently rejected. Use cases include self-tuning CNC spindles, adaptive conveyor control, predictive HVAC correction, autonomous robot re-path planning, and field-deployable OTA firmware without an engineer.
Email: info@datadoctorz.com
Landline: 044 3170 8737
Mobile and WhatsApp: +91 95510 98177
Address: Ambattur and Nungambakkam, Chennai, Tamil Nadu 600053, India
Hours: Monday to Saturday, 9:00 AM to 6:00 PM IST
DATADOCTORZ is a Chennai-based technology enterprise delivering cutting-edge solutions across Industrial Automation, Embedded AI, Machine Vision, and Legacy Data Recovery. Trusted by industry since 2008.