Digital Beamforming is one of the Eyehat corner stones, which has roots from Meharban’s experience architecting the first generation of phased array implementation for Dragon-2, the first commercial American human-rated spacecraft. Since Eyehat has infused subsequent development and simulation of similar topologies (for RF and non-RF applications) with IoT signal intelligence ethos as well as efficient, effective DSP that lends itself to robust, intuitively-tuned and metered beamforming.
From the diligence side of this technology, Eyehat has provided seminal diligence to several VC firms to help identify the true GTM potential of Hard Tech ventures exploring beamforming, as well as the key venture-specific bottlenecks to address to enable proper productionization of such systems.
Actuation Systems are another pillar of Eyehat’s engineering foundation, with several partnerships spanning Brushed, Brushless-DC Field Oriented control topology development paired with intuitive, robust tuning methods for custom and off-the-shelf controllers (on both FPGA and microcontroller targets). Meharban’s actuation system roots were established while working on Starship’s development grid-fin actuation system, thrust vector control system, as well as MEMS-scale actuation systems. Since, Eyehat has played a large part in the simulation and development of several control topologies for lunar rover actuation, actuation-based nuclear reactivity control, microgrid inverters, and coupled hydraulic-electromechanical actuation systems.
From the diligence side of this technology, Eyehat has provided diligence pertinent to where make-versus-buy trades are core to the overall viability of Hard Tech ventures in need of actuation capability. This diligence has also addressed where actuation performance makes or breaks product viability and DFM paradigms.
Sensor Topology and Digital Signal Processing touch all of Eyehat’s engineering endeavors whether it be for actuation, modem/beamformer, medtech, power systems development. Meharban spent approximately half of his career working on sensors at SpaceX ranging from transducers, speed sensors, to novel propellant estimation technology and essential navigation sensor topologies. He aims to incorporate common signal processing techniques that often come from his background in wireless communications – there is more in common between these worlds than meets the eye!
From the diligence side of this technology, similar to Actuation systems, the focus here is mainly around make-versus-buy trades with respect to monolithic sensor solutions and establishing line of sight to financially viable vertical engineering of such systems. Oftentimes, proof-of-concept is tied to single-source supply chain and procurement, and parsing the worthwhile vertical integration components is one of the key GTM figures of merit. Eyehat has also helped highlight opportunity for Software-Defined-Hardware (SDH) paradigms for several Hard Tech ventures to aid in GTM mobilization.
Position, Navigation, and Timing (PNT) systems are another key facet of Eyehat’s engineering space, and just like Digital Beamforming, embody a heavy IoT ethos philosophy to wring out as much metadata and concurrence as possible from seemingly unrelated hardware and software systems operating in the PNT application. Meharban’s work on communications systems modeling provided him trial-by-fire education on orbital mechanics, and attitude control systems. The Dragon-2 phased array algorithm development then tied the wireless communications and GNC worlds together. Since, Eyehat has developed PNT solutions which hybridize communications protocols with ranging and velocimetry for taut, precise firmware implementation and performance.
From the diligence side of this technology, Eyehat has helped identify tie-points between several nodes of products and even companies that can force multiply PNT application performance and distributed control/networking.
AI/ML is one of the more nuanced facets of Eyehat’s engineering arsenal. Meharban spearheaded several forays into ML for several SpaceX image processing and ranging-related products. As a consequence, several high-fidelity DFM trades and metering methods were established for high-precision sensor, and target tracking applications. Since, Eyehat has turned a keen eye towards augmenting AI solutions with traditional ML methods to create repeatable, precise AI/ML products ranging from structured-prompt LLM products, to in-situ tuning of signals from PNT, medtech and actuation sensing nodes.
Eyehat diligence for this technology has centered around efficient model reinforcement, and prompt efficiency. The energy and latency impact of LLMs becoming ubiquitous presents Quality of Service challenges that may not be obvious at the prototype stage, but can severely hobble the scalability of AI/ML topologies. In parallel, Eyehat has provided diligence on the longevity of AI/ML solutions – this has involved metricing the extensibility and mutability of the core AI/ML topology, given its target application and users.
ENGINEERING COVERAGE MATRIX
A systems view — not isolated disciplines.
Eyehat’s work spans tightly coupled hardware and software domains across multiple stages of technical maturity.
Rather than treat disciplines in isolation, Eyehat evaluates how architecture, sensing, control, communications, and software interact across the full FOAK → NOAK lifecycle.
The matrix below provides a high-level view of where Eyehat most commonly engages.